Post on 16-Nov-2021
Durham E-Theses
One loop phenomenology of type II string theory:
intersecting d-branes and noncommutativity
Goodsell, Mark Dayvon
How to cite:
Goodsell, Mark Dayvon (2007) One loop phenomenology of type II string theory: intersecting d-branes and
noncommutativity, Durham theses, Durham University. Available at Durham E-Theses Online:http://etheses.dur.ac.uk/2915/
Use policy
The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission orcharge, for personal research or study, educational, or not-for-pro�t purposes provided that:
• a full bibliographic reference is made to the original source
• a link is made to the metadata record in Durham E-Theses
• the full-text is not changed in any way
The full-text must not be sold in any format or medium without the formal permission of the copyright holders.
Please consult the full Durham E-Theses policy for further details.
Academic Support O�ce, Durham University, University O�ce, Old Elvet, Durham DH1 3HPe-mail: e-theses.admin@dur.ac.uk Tel: +44 0191 334 6107
http://etheses.dur.ac.uk
2
One Loop Phenomenology of Type II String Theory:
Intersecting D-Branes and N oncommutativity
Mark Dayvon Goodsell
The copyright of this thesis rests with the author or the university to which it was submitted. No quotation from it, or information derived from it may be published without the prior written consent of the author or university, and any information derived from it should be acknowledged.
A Thesis presented for the degree of
Doctor of Philosophy
• ,,Durham University
Centre for Particle Theory Department of Mathematical Sciences
. Durham University .~ England
·~ May2007
1 1 JUN 2007
One Loop Phenomenology of Type II String Theory:
Intersecting D-Branes and Noncommutativity
Mark Dayvon Goodsell
Submitted for the degree of Doctor of Philosophy
May 2007
Abstract
We examine one loop amplitudes for open and closed strings in certain D-brane
configurations, and investigate the consequences for phenomenology.
Initially we consider open strings at D6-brane intersections. We develop tech
niques for one-loop diagrams.The one-loop propagator of chiral intersection states
is calculated exactly and its finiteness is shown to be guaranteed by RR tadpole
cancellation. The result is used to demonstrate the expected softening of power law
running of Yukawa couplings at the string scale. We also develop methods to calcu
late arbitrary N-point functions at one-loop, including those without gauge bosons
in the loop. These techniques are also applicable to heterotic orbifold models.
One issue of the intersecting D6-brane models is that the Yukawa couplings of
the simpler models suffer from the so-called "rank one" problem - there is only
a single non-zero mass and no mixing. We consider the one-loop contribution of
E2-instantons to Yukawa couplings on intersecting D6-branes, and show that they
can provide a solution. In addition they have the potential to provide a geometric
explanation for the hierarchies observed in the Yukawa couplings. In order to do this
we provide the necessary quantities for instanton calculus in this class of background.
We then explore how the IR pathologies of noncommutative field theory are re
solved when the theory is realized as open strings in background B-fields: essentially,
since the IR singularities are induced by UV /IR mixing, string theory brings them
under control in much the same way as it does the UV singularities. We show that
iii
at intermediate scales (where the Seiberg-Witten limit is a good approximation) the
theory reproduces the noncommutative field theory with all the (un)usual features
such as UV /IR mixing, but that outside this regime, in the deep infra-red, the theory
flows continuously to the commutative theory and normal Wilsonian behaviour is
restored. The resulting low energy physics resembles normal commutative physics,
but with additional suppressed Lorentz violating operators. We also show that the
phenomenon of UV /IR mixing occurs for the graviton as well, with the result that, - -
in configurations where Planck's constant receives a significant one-loop correction
(for example brane-induced gravity), the distance scale below which gravity becomes
non-Newtonian can be much greater than any compact dimensions.
Declaration
I declare that no part of this thesis has been submitted elsewhere for any other de
gree or qualification. The work in this thesis is based on research carried out at the
Centre for Particle theory, Department of Mathematical Sciences, Durham Univer
sity, England, and is the result of collaboration between the author and collaborators
Steven Abel and Chong-Sun Chu, published in
• S. A. Abel and M.D. Goodsell, "Intersecting brane worlds at one loop," JHEP
0602 (2006) 049 [arXiv:hep-th/0512072].
• S. Abel, C. S. Chu and M. Goodsell, "Noncommutativity from the string per
spective: Modification of gravity at a mm without mm sized extra dimensions,"
JHEP 0611 (2006) 058 [arXiv:hep-th/0606248].
• S. A. Abel and M. D. Goodsell, "Realistic Yukawa couplings through instan
tons in intersecting brane worlds," arXiv:hep-th/0612110.
Copyright© 2007 by Mark Dayvon Goodsell.
"The copyright of this thesis rests with the author. No quotations from it should be
published without the author's prior written consent and information derived from
it should be acknowledged".
iv
Acknowledgements
This work was supported by a PPARC studentship.
I would like to thank my supervisor Dr. Steven Abel, for generous amounts of
his time, ideas, enthusiasm and support. I also thank Dr. Chong-Sun Chu for his
collaboration on the project that constitutes chapter 5.
I would like to thank Dr. Mukund Rangamani and Dr. Daniel Roggenkamp for
valuable discussions and advice, at a point of frustration while undertaking the
work of chapter 3; and also my examiners Dr. Peter Bowcock and Prof. Fernando
Quevedo, for their constructive input and interest.
I thank my erstwhile CM337 officemates Anna Lishman/Kerwick, David Leonard,
Gavin Hardman, Lisa Muller, Jennifer Richardson and Gina Titchener (and latterly
Brett Houlding and Alicia Huntriss), for their friendship and numerous coffee breaks;
and the rest of the Mathematical Sciences department. I also thank Malcolm Parks
for many Sunday lunches, and my other friends away from Durham for keeping in
touch.
I express my gratitude to the NHS in general, but specifically Miss Green (for
getting it right second time); and the nurses at Ushaw Moor practice, in particular
Rachel Doubleday /Barbaro, for their care over a difficult fifteen months.
U nquantifiable thanks must go to my parents, for their immeasurable unwavering
support, to whom I owe my interest in and capacity for investigating fundamental
physics. I also thank my yLayta xw nconwuc;;, for their assistance and generosity; my
grandmother, for among other things the cheese and ketchup sandwiches; and my
brothers Matthew and Luke.
Finally, I reserve special thanks for my partner Katherine, for her endless affec
tion, and selfless commitment.
v
Contents
Abstract ii
Declaration iv
Acknowledgements v
1 Introduction 1
2 Some Aspects of String Theory 8
2.1 Worldsheets, Actions and Backgrounds 8
2.1.1 Preliminaries 8
2.1.2 Gauge Fixing 9
2.1.3 Supersymmetry 11
2.2 Conformal Field Theory 14
2.2.1 The Operator Product Expansion 14
2.2.2 The State-Operator Mapping 16
2.2.3 The String Spectrum . . . . . 18
2.2.4 Bosonisation, Fermions and the GSO Projection 20
2.2.5 Partition Functions . 21
2.2.6 The String S-Matrix 22
2.2.7 Pictures • 0 0 0 •• 23
2.3 D-Branes and Orientifolds 23
2.3.1 R-R Charges 23
2.3.2 Gauge Theory: Chan-Paton Factors . 25
2.4 Intersecting Brane Worlds 0 ••• • • 0 ••• 26
Vl
Contents vii
2.4.1 Introduction . . . • 0 •••••••• 26
2.4.2 Mode Expansions and Quantisation 28
2.4.3 Phenomenology . . . 30
2.4.4 Tadpole Cancellation 32
2.4.5 Instanton Calculus 33
2.5 Non-Commutative Geometry. 35
3 Intersecting Brane Worlds at One Loop 38
3.1 Introduction . . . . . . . . . 38
3.2 One-Loop Scalar Propagator 40
3.2.1 Four-Point Amplitudes 40
3.2.2 Vertex Operators ... 41
3.2.3 Field Theory Behaviour 43
3.2.4 The Scalar Propagator 45
3.2.5 Factorisation on Partition Function 47
3.3 Divergences •• 0 • 0 • 0 ••••••••• 48
3.4 Running Yukawas up to the String Scale 51
3.5 Further Amplitudes 56
3.6 Conclusions 0 ••• 57
4 Realistic Yukawa Couplings through Instantons in Intersecting Brane
Worlds
4.1 Introduction
4.2 Instanton Calculus . 4.2.1 Tree Level Contributions
4.2.2 Pfaffians and Determinants
4.2.3 Vertex Operators and Zero Modes .
4.3 Yukawa Coupling Corrections ...... .
5 Noncommutativity from the string perspective: modification of
gravity at a mm without mm sized extra dimensions
59
59
62
62
62
66
67
71
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Contents
5.2 General remarks on UV /IR mixing in string theory
5.3 UV /IR mixing in the bosonic string ..
5.3.1 Brief review of planar diagrams
5.3.2 Non-planar diagrams in the k2 /ex'» 1limit
5.3.3 Non-planar diagrams in the k2 /cx' « 1limit
5.4 Supersymmetric Models ....... .
5.5 The two point function of the graviton
5.6 Phenomenology: modification of gravity at a mm
5.7 Conclusions
6 Summary
Appendix
A Material Regarding Intersecting Brane Worlds
A.1 Boundary-Changing Operators . . . . . . .
A.2 One-Loop Amplitudes With Gauge Bosons
A.2.1 Classical Part
A.2.2 Quantum Part .
A.2.3 4d Bosonic Correlators
A.3 One-Loop Amplitudes Without Gauge Bosons
A.3.1 Classical Part
A.3.2 Quantum Part .
A.3.3 Fermionic Correlators.
A.4 Theta Identities . . . . . . . .
viii
76
80
82
84
85
87
88
94
99
101
103
103
. 103
107
107
111
112
113
113
. 116
. 118
. 119
B Material Regarding Instanton Calculations in Intersecting Brane
Worlds
B.1 4d Spin Field Correlators .
B.2 Spin-Structure Summation
C Field Theory Limits of String Diagrams - a Review
Bibliography
120
120
122
124
126
List of Figures
3.1 An annulus contribution to 3 point functions. . . . . . . . . . . . . . 39
3.2 Feynman diagrams in the field theory equivalent to our limit; we have
factorised onto the scalar propagator, and consider only gauge and
self-couplings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.3 A linear plot of the "running" Yukawa coupling in modular parameter
t, lower graph. The peak is very close to t = 1, i.e. the string scale.
The top graph is the standard power-law behaviour continued. The
middle graph is the field theory approximation using string improved
propagators as defined in the text. . . . . . . . . . . . . . . . . . . . . 55
4.1 Contributions to 3 point functions: tree level and one-loop instanton. 61
IX
Chapter 1
Introduction
String theory is a research programme built upon the almost Aristotelian attempt
to deduce the fundamental laws of nature through reasoning; it starts from a few
basic premises and arrives at a very general and beautiful theory with many startling
properties. However, in the modern scientific world beauty is not the arbiter of the
success of a theory: only experiment has that privilege, and thus the goal of string
theorists is to connect it to the Standard Model, the most successful and accurate
theory ever created.
The Standard Model of particle physics is built on Quantum Field Theory, in
which every form of matter is associated with a field operator, and is at once a wave
and a point-like particle. Infinities in calculations which must be absorbed into
renormalisation constants are interpreted as signalling the breakdown of the theory
at some currently inaccessible energy: it is built in that it is an incomplete descrip
tion of nature, even without gravity. When we consider gravity, the scale at which
the theory must break down becomes apparent; through considering the constants
of the theories of particles and gravitation we find the Planck scale, 1.22 x 1019GeV,
to be significant. Since Quantum Field Theory is incompatible with the theory of
gravity, General Relativity, we assume that this must be the energy at which we
would certainly see evidence of the theory that provides a true description of both.
We have a clear theoretical signal that the theories need to be modified, although
essentially no signal as to how this should be accomplished, and unfortunately the
energy at which we would definitely find the new theory is prohibitively beyond
1
Chapter 1. Introduction 2
current experimental technology.
Since Quantum Field Theory is a general framework, its unification with General
Relativity is not the only demand that we place upon our new theory. We require it
to explain or predict at least some of the properties of the Standard Model, which
consists of a curious pattern of fields: three families of fermions with a hierarchy of
masses; each family having a pair of quarks and a pair of leptons, one of which is
virtually massless; the bosons carrying the gauge forces which bind matter together
with the strong and electroweak coupling, and the Higgs, the field whose vacuum ex
pectation value breaks the electroweak symmetry and provides (almost) everything
with mass. We also demand superpartners for all of these fields and a mechanism of
supersymmetry breaking, in order to account for the disparity in hierarchy between
electroweak and Planck energy scales (or possibly some other explanation). A pre
diction of these properties would also illuminate some of the cosmological problems,
such as the existence of Dark Matter (widely assumed to be the lightest superpart
ner of the Standard Model fields), although we would also desire an explanation of
Dark Energy and the apparent non-vanishing cosmological constant.
String theory provides a tantalising route to progress on all of the above prob
lems. By assuming that the fundamental objects are strings rather than particles,
it is possible to unify Quantum Field Theory with General Relativity; the appear
ance of the Einstein-Hilbert action as a coherent state of gravitons is a beautiful
and surprising result. However, without supersymmetry we find that string theory
is inconsistent: purely bosonic string theory contains tachyons1 , and whether these
condense (and what they condense to) is an open problem. Thus the programme is
intimately tied in with a supersymmetric explanation of the Hierarchy Problem.
The analogue in string theory of the choice of fields of the Standard Model is the
background of string compactification. The natural setting for string theory is fiat
ten-dimensional spacetime, and thus we must assume instead that (usually) six di
mensions are compact and small enough to be currently undetected. The properties
of the compact dimensions then dictate the spectrum of fields observed by macro-
1 Although bosonic string theory does provide a useful toy framework, a fact relied upon in this
thesis.
Chapter 1. Introduction 3
scopic observers. Initially it was hoped that consistency would dictate an essentially
unique theory; however, without further theoretical constraints the current under
standing is of a presumably infinite landscape of possible backgrounds. This can be
regarded as similar to the possibilities of fields and gauge groups in the framework
of Quantum Field Theory. A particular non-trivial enterprise in string theory is the
construction of backgrounds which reproduce the Standard Model (or supersymmet
ric extensions thereof) at low energies, and although it is no longer expected that a
unique example will be distinguished from the many other examples, it is valuable
to find the features of such edifices that account for the various mysteries of the
Standard Model. Generally features of the Standard Model have simple geometric
interpretations in string constructions, although some are rather subtle. It is then
hoped that the model-building techniques will reveal the necessary theoretical input
to provide an ultimate theory.
The current understanding of String Theory is that it is one theory with several
formulations; in ten dimensions it is described by type I, IIA, liB and heterotic
S0(32) and E 8 x E 8 :
• Heterotic string theory was long considered the main candidate for construct
ing realistic models. It consists of closed strings where the oscillations propa
gating in opposite directions around the loop experience different dimensions
of spacetime with different amounts of supersymmetry; while the right moving
modes experience supersymmetry amongst the bosonic and fermionic degrees
of freedom and propagate in ten dimensions, the left moving modes have no
supersymmetry and feel twenty-six dimensions. Ten of the left-moving dimen
sions are considered to be bosonic spacetime coordinates, while the remaining
sixteen are converted into thirty- two real fermionic degrees of freedom, ac
counting for the internal symmetry S0(32) or E 8 x E 8 (these specific groups
are required for the cancellation of anomalies). Starting from such a large
symmetry group, model building involves specifying a fibration of the internal
degrees of freedom over the six compact spacial dimensions in such a way as to
break the symmetry, to obtain grand unified or standard model gauge groups
and flavours. This was an attractive proposal, as it offers a "top down" ap-
Chapter 1. Introduction 4
proach where the fundamental theory is specified and "all" that is required is
a manifold. There is still much work in this area, but we shall not consider it
further.
• Type I strings are unoriented, in that they are invariant under exchange of
left- and right-moving modes of closed strings, and contain open strings whose
endpoints are indistinguishable. The endpoints of open strings do, however,
must have gauge degrees of freedom, which anomaly cancellation determines
to be 80(32).
• Type II strings were originally purely closed strings, with supersymmetry for
both left- and right-moving modes, and hence propagating in ten dimensions
with no additional internal degrees of freedom. This made them initially
unattractive for model building until the advent of D-branes.
All of the above theories are related by dualities, and the overarching theory is
M-theory, which exists in eleven dimensions and yields the other theories upon com
pactification; however, its formulation is unclear or difficult to work with (the Matrix
Theory proposal is a subject of much interest, but great difficulty for model build
ing). Hence string phenomenology predominantly focuses on the ten-dimensional
theories. They are all defined in terms of a perturbative expansion, and the non
perturbative definition is not known. However, several non-perturbative objects
have been found. Upon a transformation of the theory known as T-duality, D
branes (so called because they are hypersurfaces with Dirichlet boundary conditions
for the strings) appear, the excitations of which are string states, and as such pro
vide important tools in model building. By stacking several branes together, the
string endpoints obtain gauge degrees of freedom, and thus type II strings attached
to D-branes have become subject to significant recent interest; indeed this thesis
shall focus upon that framework. Related to these are NS-5-branes, six-dimensional
hypersurfaces upon which D-branes may end, and Orientifold Planes, which are non
dynamical hypersurfaces left invariant under an orientation reversal of the strings
coupled to a reflection in space; the latter of these two shall also be relevant to this
work.
Chapter 1. Introduction 5
D-branes offer an elegant way to restrict matter fields to a four-dimensional hy
persurface, and hence have provided an important tool in model building. However,
since gravity is mediated by closed strings not restricted to the branes, and since
we observe four-dimensional gravity and no gravitinos, string models generally com
pactify six dimensions and break (at least some of) the supersymmetry in them.
Hence some approaches involve just a D3-brane (3 labelling the number of spatial
dimensions) in either a complicated Calabi- Yau manifold (a manifold with vanish
ing Ricci tensor and having SU(3) holonomy, necessary and sufficient to be a string
background with N=l supersymmetry) or a simple subset thereof, an Orbifold: a
quotient of flat space by a subgroup of its symmetries. We shall have these models in
mind in chapter 5, but will not refer to the specific details of them there. There are
other approaches, where our four dimensions are considered to be the intersection of
two D-branes for some fields. Specifically, a very popular construction involves sev
eral sets of D6-branes, where chiral superfields exist at the intersections and vector
superfields are found on the full volume of the branes. These shall be of particular
relevance in chapters 3 and 4. Since in these models we must always compactify the
extra dimensions, and we wish to break supersymmetry for gravity, the same tech
niques involving orbifolds or other manifolds apply in each case. Indeed, they can
usually be related by dualities, and the D3-brane models are generally augmented by
the introduction of D7-branes, often required for consistency. However, the case of
a toroidal orientifold background (essentially flat space with every extra dimension
periodic, and a choice of orbifold and orientifold group) has attracted much interest
since it allows many quantities to be calculated exactly while having sufficiently
realistic properties to provide useful toy models. It is these models that we shall
consider. There has been considerable interest in further embellishing these mod
els by adding certain types of closed string field backgrounds (so called three-form
fluxes) in order to stabilise the size and shape of the compact dimensions against
deformation whilst breaking supersymmetry. However, upon introduction of these
fluxes most quantities become (currently) incalculable in full string theory and it is
only possible to apply supergravity or field theory analysis. Hence we shall neglect
these models.
Chapter 1. Introduction 6
There remain several challenges for string phenomenology. Overall the aim is to
provide a fully realistic model of the real world with a convincing method of super
symmetry breaking; it would then be hoped that there would be some predictive
relation between the parameters. Working toward this within the sub-field of D
brane constructions, the low energy properties are still not completely understood;
we consider N = 1 supersymmetric models, for which the perturbative supeq)o
tential receives no corrections at one loop, but the remaining properties (such as
running of the couplings) are determined by the Kahler potential, which is poorly
understood. The work in chapter 3 adds to the effort to understand it. It investi
gates the appearance of divergences that would render the theory inconsistent, and
their cancellation. It also outlines new techniques for the calculation of diagrams
necessary for the determination of the Kahler potential, and determination of the
running of Yukawa couplings.
Although the superpotential is dominated by a perturbative component that is
determined completely at tree level in these models, there are also non-perturbative
contributions involving both tree and one-loop amplitudes, and they have not pre
viously been calculated. The work of chapter 4 provides the essential tools for the
calculation of these instanton contributions, and discusses how they can solve certain
phenomenological problems in such frameworks.
Another challenge is to relate cosmological data or possible experiments to our
particle models. In the study of D-branes it was discovered that string theory, under
certain circumstances, gives rise to noncommutative geometry. This is a phenomenon
where the operators corresponding to different coordinate directions have a non
zero commutator. It is an area of considerable work at present, with some hoping
that it will provide a new method to unify field theory and gravity. From the
phenomenologist's perspective, it provides many intriguing new behaviour in field
theory, such as UV /IR mixing, where low energy interactions actually probe the
high energy behaviour of the theory. One occasion that it occurs in string theory is
the relative motion of two D-branes, and can be viewed as an uncertainty relation;
an analogue of the Heisenberg relation. In this way, it is supposed by many to play
a key role in the ultimate theory, and thus merits considerable current research. Its
Chapter 1. Introduction 7
appearance would have many consequences for phenomenology, and would give clear
signals in experiments; Chapter 5 investigates some of these in a framework where
it appears naturally in string theory, including hitherto unexplored consequences for
gravity.
We turn now to chapter 2, which provides an introduction to many of the con
cepts required for the rest of the thesis, including the formulation of type II string
theory and a discussion of D-branes.
Chapter 2
Some Aspects of String Theory
This chapter briefly reviews the aspects of string theory relevant to the subsequent
chapters, providing necessary references. For more material on the subject the reader
should consult [1, 2].
2.1 Worldsheets, Actions and Backgrounds
2.1.1 Preliminaries
In the path-integral formulation of field theory, the fundamental objects (represent
ing particles) are fields which may take any value at any point in space, and the
amplitude for transition from one point in space (or momentum space) to another is
given by integrating over all paths with a weight given by the action for that path or
worlclline. In string theory, the fundamental objects are one-dimensional and thus
sweep out a worlclsheet, so the path integral is weighted by an action depending
on two variables. In bosonic string theory, the fields in the theory are then the
spacetime coordinates, while in the superstring we also introduce fermionic coordi
nates for each spacetime dimension. We have an additional choice to make, that is
whether the worldsheet has a boundary- this determines whether the fundamental
objects. are closed or open strings.
To study the theory, we must begin with an action. Without boundary terms
(which we shall return to later), the purely bosonic part of the action for a general
8
2.1. Worldsheets, Actions and Backgrounds 9
spacetime is taken to be a nonlinear sigma model and can be written
Sx = 4:a' JM d2ah112 [ (habGttv(X) + icab Bttv(X)) 8aXIlEJbXv +a' R<I>(X)]
(2.1.1)
where hab is the Euclidean worldsheet metric (in two dimensions, even with gravity,
the Wick rotation from a Minkowski worldsheet is well justified) and { a 1 , a 2 } are the
worldsheet coordinates - generally we consider a 1 to parametrise the direction along
the string, and a 2 to be the (Euclidean) time; Xtt are the target space coordinates;
Gttv is the Minkowski spacetime metric; Bttv is an antisymmetric tensor field; R is the
curvature of the worldsheet and <I> is the dilaton field. The dimensionful constant a'
is proportional to the string length squared, and sets the scale of string interactions:
as we take a' ---+ 0, we expect to obtain particle behaviour. The label M in the
above action denotes the worldsheet geometry, which we must specify.
The fields G, B and <I> are all considered to be (potentially running) couplings
m the quantum theory - in order for string theory to be consistent on a given
background spacetime it must be stable to quantum corrections of these fields, i.e.
their beta-functions must vanish. It is reassuring to note that we can take flat
space to be a consistent background; the general conditions are irrelevant for our
purposes, but the backgrounds we shall be interested in are Ricci-flat (Rab = 0) and
thus we may consider six of the dimensions to be a compact manifold provided we
maintain Ricci-flatness. The backgrounds of interest to us will have constant (or
zero) antisymmetric tensor, which plays an important role when we consider open
strings (worldsheets with boundaries). They also have a constant dilaton <I>0 , which
plays a vital role: it becomes the coupling constant of the theory, since the term
J R = x is topological and so assigns a coupling to worldsheets of different Euler
characteristic X· The path integral is then split into a sum over Euler characteristics,
which is interpreted as a sum over numbers of loops.
2.1.2 Gauge Fixing
The action (2.1.1) was chosen to satisfy the symmetries that we expect: those of
the spacetime background, and diffeomorphisms of the worldsheet. However, it
possesses an additional worldsheet symmetry: Weyl invariance. It is invariant under
2.1. Worldsheets, Actions and Backgrounds 10
local rescaling of the worldsheet metric
(2.1.2)
and thus is (at least so far classically) a conformal field theory. In creating a quantum
theory of strings, we write a path integral using the above action, and we integrate
over all possible worldsheets and embeddings into the spacetime coordinates. How
ever, since we have diffeomorphism and also Weyl invariance of the worldsheet, we
will have an inconsistent path integral unless we divide the measure by the volume
of this symmetry group. This can be achieved using light-cone coordinates, but the
covariant approach is to use the Faddeev-Poppov procedure and introduce ghost
fields to fix the gauge. We write the Polyakov path integral
J [dXdh] ( ... ) = L Vol( G)( ... ) exp[-S- ~ox]
X
(2.1.3)
where G is the product of diffeomorphism and Weyl symmetry groups, and ... repre
sents an insertion of operators; and insert the Faddeev-Poppov measure in the usual
manner. The result is that the integration over metrics is reduced to an integral over
moduli (parameters of the metric which can not be removed through diffeomorphism
and Weyl transformations) and we can fix the worldsheet metric to be conformally
flat:
(2.1.4)
but the price that we pay is the inclusion of anticommuting ghost fields b, c and
b, c. Each modulus of the worldsheet metric contributes a b-ghost insertion; for our
purposes, we shall only need consider the topologies of the disk, torus and annulus.
The disk has no moduli, the annulus has one, and the torus has two. The ghost
insertions for the annulus and torus are respectively 27rib(O) and -47rb(O)b(O).
With the new flat metric, we choose complex coordinates
(2.1.5)
and write the ghost action as
(2.1.6)
2.1. Worldsheets, Actions and Backgrounds 11
In addition, there are still symmetries not fixed by specifying the metric, that
must fixed by the Faddeev-Poppov procedure: these comprise the conformal Killing
group (CKG), which is a subgroup of G. The generators of this group are called the
conformal Killing vectors ( CKV s), and for each of these we must fix a coordinate
of an inserted operator (consistent with the CKG) and insert a c-ghost. The ghost
insertions are independent of the target space geometry - they depend only on the
class of worldsheet - and so the ghost portion of the path integral is universal to
all string scattering amplitudes. Of relevance here is the result for the annulus: the
determinant of the ghost insertions is given by 1'17(it)l 2, where tis the modulus of
the annulus such that the fundamental domain is taken to be [0, 1/2] x [0, it] and 17
is the Dedekind Eta function.
2.1.3 Supersymmetry
We now wish to consider adding fermions to the theory. This is accomplished by
implementing worldsheet supersymmetry. For a general background, the non-linear
sigma model is complicated, so we shall specialise to the case of flat space. Then we
can write the action, in conformal gauge, as
(2.1.7)
where 'lj;IL and {j;~.t are Grassmann ( anticommuting) fields. We may write the stress
energy tensor associated with variation of the worldsheet metric
(2.1.8)
which is conserved classically, translating to Tww _ T 8 being holomorphic, and
T1111v = T8 being antiholomorphic. Similarly, we have a tensor associated with
superconformal transformations, with respectively holomorphic and antiholomorphic
2.1. Worldsheets, Actions and Backgrounds
components TF ( w) and f'F ( w):
( 2) 1/2 i - 'lj; 118X
a' 11
( 2) 1/2 - -i - 'lj; 118X
a' 11
12
(2.1.9)
If we look for the classical equations of motion from the above action, we first
vary X and obtain
where 8 M is the boundary of the worldsheet M. This leads to the equation of
motion:
(2.1.11)
and thus classically X 11 splits into holomorphic and antiholomorphic parts. It also
gives us the boundary conditions: if we consider the string to be a closed loop, then
the worldsheet has no boundary (and clearly we must consider it periodic around
the "length" direction a 1), but if it is an open string then, since we consider the
(}'2 direction to be the (Euclidean) time, the lowest genus worldsheet is an infinite
strip, which we shall take to have boundaries at ~(w) = 0 and ~(w) = 1/2, and a
possible boundary condition is
(8 + B)X 11 = o, ~(w) ~ 0, 1/2. (2.1.12)
These are Neumann boundary conditions. The alternative is to consider the string
endpoints to be fixed at the boundaries:
~(w) = 0, 1/2 (2.1.13)
which corresponds to
(8-EJ)X=O, ~(w) = 0, 1/2 (2.1.14)
since this is the derivative along the "time" direction. These are Dirichlet boundary
conditions, the significance of which we shall return to later.
2.1. Worldsheets, Actions and Backgrounds 13
Considering now variations of the fermionic coordinates, we obtain
0 = 6S = ~ f d2w6'1j;jjfh/Jjj+6;f;j3;f;jj+4i j (;f;jj6;j;jj)dw-('lj;jj6'1j;jj)dw. (2.1.15)
21r J M 1f laM
The equations of motion are therefore
(2.1.16)
and thus axjj and 1/Jjj are holomorphic fields, and axjj and ;j;jj are antiholomorphic.
The boundary conditions for open string fermions on the infinite strip are
~(w) = 0, 1/2 (2.1.17)
which leads to
;f;jj(w), ~(w) = 0
±;f;jj(w), ~(w) = 1/2. (2.1.18)
We can use the "doubling trick" to combine these fields into one holomorphic field
on an infinite cylinder, of domain [-1/2, 1/2] x [-oo, oo]:
{
'lj;P(w) ~(w) > 0 1jJP(w) = -
1jJP(-w) ~(w) < 0 (2.1.19)
This converts the boundary conditions to a periodicity: 1/Jjj(w + 1) = ±1/Jjj(w) =
e2niv'I/Jjj(w). The choice of a positive sign (v = 0) is :ailed Ramond (R) bound
ary conditions, and a negative one (v = 1/2) Neveu-Schwarz (NS). Closed string
fermions may also be periodic or antiperiodic, but the 1jJP and 'lj;P fields have this
condition separately and thus have four sectors: R-R,R-NS,NS-R and NS-NS (in
principle, the bosons could also be antiperiodic, but this would be inconsistent with
the symmetry of the target space).
Having established the equations of motion and boundary conditions, we can
write down mode expansions for the fields as free waves. We shall be primarily
interested in open strings, for which we can write
8X"(w) = -i ( ~') I/2 f a~e-iw(k+I) k=-00
(2.1.20) k=-oo
2.2. Conformal Field Theory 14
When we quantise the theory, the coefficients become operators; the normalisation
of the above is a choice that gives the (anti)commutators of the coefficients integer
values, given by
J.lV ~ ffiTJ Um+n
(2.1.21)
Note that the bosonic and Ramond fields have zero-modes; the bosonic ones corre
spond to the centre-of-mass momentum, a~= (2a') 112pJ.L. The Ramond zero modes
will allow for spacetime fermions.
Finally in this section, having introduced the superpartners of the bosonic fields,
in gauge-fixing the action we also require the superpartners of the ghost fields.
Since the original ghosts are Grassmann fields, we find that we have commuting
fields {3, 'Y, {3 and ;y, with action
1 J 2 -Ssg = 2
7!" d w{3[}"( (2.1.22)
and conjugate for the antiholomorphic {3 and ,:Y; for open string the antiholomor
phic fields can be combined with the holomorphic ones using the doubling trick as
elsewhere.
2.2 Conformal Field Theory
We now consider the quantum theory based on the actions considered in the previous
section.
2.2.1 The Operator Product Expansion
If we have a set of local operators ~Pi(z, z) in a quantum theory, then the operator
product expansion states that we can express the product of two of these operators
as a sum over operators at a point:
~Pi(z, z)IPJ(w, w) = L c7J(z- w)~Pk(w, w). k
(2.2.1)
As an operator equation, this holds inside a general correlator provided that I z-w I is smaller than the distance to any other operator insertions. Since the string action has
2.2. Conformal Field Theory 15
conformal symmetry, we expect this to affect the operators of the theory, and indeed
we find operators of interest have definite weights under global transformations:
(2.2.2)
where ,\ is a complex number and (hi, hi) are the conformal weights of <I>i. This
property is actually required for unitarity of the theory, and restricts the OPE to
<I>i(z, z)<I>j(w, w) = I)z- w)hk-h;-hj (z- w)iik-h;-hjc~j<I>k(w, w). k
(2.2.3)
so that now the cfj are constants. In fact, all the properties of the theory can be
extracted from those of the primary fields, which transform as
(2.2.4)
The OPEs of primary fields generally have only finitely many singular terms, which
determine most of the properties. For the string action, we can determine the OPEs
by examining the path integral [1] or by considering the correlator of two fields. The
result is 8X11-(z)axv(w) a' TJ~'v b(z)c(w) 1
rv 2 (z-w)2 rv (z-w)
tJX11-(z)axv(w) rv a'~ b(z)c(w) 1 2 (z-w) 2 rv (z-w) (2.2.5)
1jJ11-(z )'1/f( w) rv~ f3(zh(w) -1 z-w rv (z-w)
{/;11- ( z)?/f ( w) rv~ /1(z)i(w) -1 z-w rv (z-w)'
where rv denotes only the singular terms in the OPE. From these, we can define a
conformal normal ordered product of pairs of operators: <I>i(z)<I>j ( w) : by subtracting
the singular terms in the OPE, and thus have a consistent definition of coincident
operators - also giving a product which obeys the classical equations of motion.
Thus the stress-energy tensor is written
where D is the number of spacetime dimensions. As we noted, the stress-energy
tensor is the Noether current associated with variations of the metric, and thus is
associated with conformal transformations. In the quantum theory, this becomes
2.2. Conformal Field Theory 16
a Ward identity, with an infinite number of currents. The Ward identity gives
the transformation of operators when we make a conformal transformation, and
determines the OPE of the stress-energy tensor with a primary field <I>(w, w) to be
B _ h _ 1 _ T (z)<I>(w, w) r-v (z _ w) 2 <I>(w, w) + (z _ w) 8<I>(w, w). (2.2.7)
From the action we can see that the weights of 8XJ.l, fJXJ.l, 'lj;J.l and ;j;J.l must be (1, 0),
(0, 1), (1/2, 0) and (0, 1/2) respectively, and this can also be seen from the OPEs,
the definition of the stress-energy tensor and the above equation.
We also have a stress-energy tensor for the ghosts and superconformal ghosts,
with components T9 (z), T89 given by
: b(z)8c(z) : -28(: b(z)c(z)) : 3
: 8{3(z)'y: - 28(: f3(z)'y(z) :) (2.2.8)
and Tf(z), T8~(z) given by the conjugate. The coefficients 2 and ~ in the second
terms are determined by the Faddeev-Poppov procedure (the above should be com
pared with the 'lj;J.l stress-energy tensor) which fixes the conformal weights to be
(2, 0), ( -1, 0), (0, 2) and (0, -1) for b, c, band c respectively, and (3/2, 0), ( -1/2, 0),
(0, 3/2) and (0, -1/2) for {3, "(, /3 and ;y.
If we now consider the 0 PE of the stress-energy tensor with itself, we find that
in general it is not a primary field:
B B C 2 B 1 B T (z)T (w)"' 2(z- w)4) + (z- w)2T (w) + (z- w) 8T (w) (2.2.9)
where c is the central charge, which for our case after summing over the stress-energy
tensors for ax 1 'lj;' ( b, c) and ({3 1 'Y) is given by
cTot = ~D- 15. 2
(2.2.10)
It transpires that this must be zero to have a consistent BRST operator, and for the
Weyl symmetry to not be anomalous in the path integral: we thus conclude that
there must be 10 fiat spacetime dimensions.
2.2.2 The State-Operator Mapping
Having established the properties of the operators in our conformal field theory, we
now wish to make the connection with states in our string Hilbert space. This is
2.2. Conformal Field Theory 17
done via equation (2.1.20). We first consider fields on the infinite cylinder, and make
a conformal transformation to turn this to the complex plane, using the variable
z = -e-inw. This then translates the infinite past SS(z) = -oo to the origin, and
thus the asymptotic state of a field on the cylinder is found by considering it at the
origin. We define
lci>in) =lim ci>(z, z)IO) z--->0
(2.2.11)
and thus if we expand 4> as a mode expansion, the positive frequency modes must
annihilate the vacuum:
m,n
4>m,n IO) = 0, m > -h,n >-h. (2.2.12)
If we take the Hermitian conjugate of a field, then, since we are using Euclidean
time, we must make the replacement z---+ 1/z, and demanding that the asymptotic
out state has a well-defined inner product with the in state requires
(2.2.13)
which translates to the condition on the mode operators
(2.2.14)
and hence the negative modes are considered to be creation operators.
Note that in the variable z, we have a Laurent expansion, and so we can perform
a contour integral to invert it:
4> =- dzzm+h-1_ d.z.zm+h-lq>(z z). 1 f 1 f -m,n 27ri 2Jri ' (2.2.15)
So we have
"~miO);n ~ ( ~' f' ( m ~ l)! iJ"' X"(O), m>O. (2.2.16)
This is the State- Operator Mapping. For the bosonic fields we find that there is
no additional operator associated with the vacuum - it is represented by the unit
operator. However, in general we find that we need an additional operator for the
vacuum.
2.2. Conformal Field Theory 18
2.2.3 The String Spectrum
In the operator formalism, we have the concept of conformal normal ordering, but in
the canonical quantisation formalism, we also have the concept of normal ordering:
the annihilation operators are placed to the right of the creation operators. It turns
out that these are related. If we consider the mode expansion of the stress-energy
tensor, ()()
T 8 (z) = L z-m-2 Lm (2.2.17) m=-oo
(the Lm are called the Virasoro generators), then from equations (2.2.6) and (2.2.8)
we can write the Lm in terms of normal ordered pairs of mode operators, except
with a constant associated with the zero mode. We find
n=-oo
L"/n ~ f (2k + 2v- m)'ljJ~-k-v'I/JJLk+v + ~ (1 - 4v2
)5m
£89 = m
k=-oo 00
n=-oo
1 ()() 1 - ~ (m + k + v)f3m-k-vrk+v + Om-(1 + 2v)(3- 2v) 2 ~ 8
k=-oo
(2.2.18)
where the above are assumed to be creation-annihilation ordered. These Virasoro
generators all separately generate the Virasoro algebra, which has a central charge
that cancels when all sectors of the theory are summed. The £ 0 operator is the
Hamiltonian of the system, and all physical states must satisfy the constraint
(2.2.19)
In the Old Covariant Quantisation scheme, this arises by requiring that the equa
tion of motion T:£ot = 0 holds as an operator equation; in BRST quantisation, it
arises from insisting that physical states are annihilated by the BRST operator con
structed from gauge-fixing the metric, and the additional requirement that they are
annihilated by the zero modes of the b-ghosts. This requirement determines the
string spectrum, since n:b = (2n:') 112k11 (where kJL is the centre of mass momentum)
we have an equation relating the rest mass of the state to the number of excitations.
2.2. Conformal Field Theory 19
Notably, we find that the NS vacuum state does not have zero rest mass:
(2o/k2- 1/2)10) = 0, NS
R, (2.2.20)
and thus we find that the lowest NS state is a tachyon. Another feature is that if we
excite the vacuum with the operator '1/Jf{ in the R sector, we do not change the rest
mass. This second property actually tells us that the Ramond states are fermionic:
the '1/Jf; operators take on the role of gamma-matrices, and form a representation of
the Clifford algebra in 10 dimensions. It is the 32 representation, which admits a
decomposition by chirality into 16 + 16'. We can choose a helicity basis by writing
_1 ('1/Jl ± .t,O) J2 0 '+'0
_1_(.1,2a ± i·'·2a+l) J2 'f'O 'f'O ' a= 1...4 (2.2.21)
and defining a state lxo) such that rn-lxo) = 0, so that we can build all other states
as
I so, St, s,, s,, s,) ~ (ft (r"+)•,.+l/') lxo), (2.2.22)
where sn = ± 1/2. Similarly we can complexify the fermion operators { 'ljJJ.L} ~
The above represent basis spinors; a physical state requires a wavefunction built
using these, so we can write, for example, ua.Sa. for a left-handed fermion where Sa.
consists only of left-handed basis spinors. Similarly, the massless NS states are given
by 'l/Ji;2 IO), J-L = 0 .. 9, and so we write a wavefunction as eJ.L'l/Ji;2 IO). The equations of
motion for the wavefunctions are determined by conditions similar to the Virasoro
constraints above: they are derived from the physical state conditions arising from
the superconformal current. Writing ()()
TF(z) = L Gk+vZ-k-v-312 (2.2.23) k=-oo
we impose the constraints
r 2: 0. (2.2.24)
This gives the Dirac equation k · fu = 0 in the Ramond sector, and the transversality
condition k · e of a gauge boson in the Neveu-Schwarz.
2.2. Conformal Field Theory 20
2.2.4 Bosonisation, Fermions and the GSO Projection
Through an interesting property of conformal field theory, it is actually possible to
represent the fermionic operators in terms of bosonic ones. By considering a bosonic
field H(z), with OPE
we can write
H(z)H(w) rv -log(z- w)
wn(z)
~n(z)
: wn~n(z) :
rt
: e-iHn(z) :
i8Hn(z) 1 2 : 8Hn8Hn(z) :
(2.2.25)
(2.2.26)
since the OPEs are the same. This is bosonisation, and we can find a similar rela
tionship for the superconformal ghosts that appear in amplitudes using a field </>(z)
with "wrong" sign OPE </>(z)</>(w) rv log(z- w). If we examine equation (2.2.18),
we find that the correct weight for the vacuum state is only possible if we include
an operator e1<P, with l = -1/2 for the Ramond ground state, and l = -1 for the
Neveu-Schwarz. l is referred to as the </>-charge of the state. In addition, from the
above equations we can see that we can write the Ramond ground state under the
state-operator mapping as
is0 , s1, s2, s3, s4) rv e-<P/2 exp [it siHil·
n=O
(2.2.27)
The coefficients Sn now become denoted H-charge. For general (non ground state,
massive) Ramond operators, we will vary the Sn by integer coefficients by including
various wn or ~n.
We previously noted that the theory has a tachyon, and we have massless
fermions of both chiralities (yet we would like a chiral theory). Both problems
can be solved at once, by the GSO projection. This involves defining an operator F
to represent the world-sheet spinor number, and projecting out states according to
their eigenvalue under exp(niF). A formal definition ofF can be made in terms of
the spinor number currents : Wn ~n(z) : and : {Jy(z) : and requiring mutual locality
2.2. Conformal Field Theory 21
(integer-moded OPE) with the BRST operator, but we shall simply note that for
our purposes F is given by 4
F = l + LSn. (2.2.28) n=O
Since the NS vacuum is tachyonic, we wish to exclude it, and so we insist that
exp(niF)Ns = 1 for a consistent theory. However, in the Ramond sector, we can
choose either value. Indeed, we note that the action on the massless states is actually
a chirality projection, and so we obtain a chiral theory with N = 1 supersymme
try in 10 dimensions (it is possible to obtain the spacetime supercharges in terms
of the worldsheet quantities to demonstrate that the fermions and bosons are in
deed superpartners). For closed strings, we can even choose different projections for
holomorphic and antiholomorphic modes, and indeed type I I B string theory has
exp( niF) = 1 for all sectors, while I I A has exp( niF) = -1 only for the antiholo
morphic Ramond sector. The choice of GSO projection also distinguishes between
type I and type I' open string theories.
2.2.5 Partition Functions
Having established the string spectrum, we can calculate a trace over it weighted by
the Hamiltonian called the partition function. In statistical mechanics the partition
function determines many of the thermodynamic properties of a system; in string
theory it plays a similarly central role in determining many of the properties of the
theory at one loop, even before we consider adding external states. For open strings,
it is defined as
100 dt
Z = 0 2
t (Avach, (2.2.29)
where the subscript 1 denotes the one-loop worldsheet, in this case the annulus,
which has one modulus, t. Avac is the vacuum operator: in the NS sector it is
merely b0 c0 (zero modes of the ghosts), while in the R sector it must include a spin
field. To calculate the above, we could perform a path integral; however, the result
is also given by the Coleman-Weinberg formula
(2.2.30)
2.2. Conformal Field Theory 22
where F is the space-time fermion number. Since we have separate NS and R sectors
to consider, we might expect two partition functions; however, since we must impose
the GSO projection it is more convenient to separate into four as follows:
Z1(t) trR((-1t exp[-2ntLo])
Z2(t) tr R( exp[-2ntL0])
Z3 (t) trNs( exp[ -2ntL0])
Z4(t) trNs(( -1)F exp[-2ntL0]). (2.2.31)
where F is the world-sheet fermion number as before, and then write
100 dt
Z = - L bvZv(t), 0 2t
v
(2.2.32)
with phases 8v = {1, -1, 1, -1}. Finally, it is worth noting that the annulus affords
dual interpretations: either as open strings with the endpoints having specified
boundary conditions, or as closed strings propagating as a cylinder. In the latter
case, it is convenient to work in the variable l = 1/t where high energy open strings
are interpreted as low energy closed strings propagating for long distances.
2.2.6 The String S-Matrix
As previously mentioned, expectation values of groups of operators are calculated
from a gauge-fixed path integral, and a single-string initial state on a tree-level world
sheet can be mapped to an operator at the origin. However, we wish to calculate
amplitudes with asymptotic states including several strings. This is done by sewing
tree-level worldsheets containing the separate string states onto the arbitrary-loop
worldsheet. Since tree-level worldsheets have no moduli, and each contains only an
operator at its origin, its size is irrelevant and we can shrink it to a point. However,
for each worldsheet we sew in, we add a boundary for an open string, or a loop
for a closed string, albeit at infinity and of vanishing size. These change the Euler
characteristic of the surface, so although we do not increase the number of moduli
we must add a factor of e<~>o for an open string, and e2<I>o for a closed string; these
play the role of the coupling, thus we label 9s = e2<I>o, and 9o for the open string
factor. The only remaining degree of freedom is to integrate the operators over
2.3. D-Branes and Orientifolds 23
the worldsheet boundaries (for open strings) or area (for closed strings), which we
must do to ensure consistency, so finally expectation values that we calculate consist
of operators from the state-operator mapping with an appropriate coupling factor
each integrated over the worldsheet: the complete operator for a physical state in
the string 8-matrix is known as a vertex operator.
2.2. 7 Pictures
Although we integrate vertex operators over the worldsheet, due to the gauge-fixing
we must fix one coordinate and include 'a c-ghost for each CKV. Once we have
worldsheet supersymmetry we promote the Riemann surfaces in the integrals to
Super-Riemann surfaces, and thus must add superconformal ghost operators. The
number of superghost insertions required depends upon the genus just as the num
ber of c-insertions does; recalling that the superghost insertions are given by the el¢
terms, we require a total ¢-charge of 2g- 2, applying separately to holomorphic and
antiholomorphic sectors for closed strings. Hence, if we want to have an arbitrary
number of vertex operators, we must insert additional operators to give the correct
total ¢-charge; these are Picture Changing Operators (PCOs). The vertex opera
tors from the state-operator mapping are regarded as having the superpartners of
the coordinates fixed, while the picture-changed operators are equivalent to having
unfixed coordinates. The picture-changed operator is given (for the purposes of this
thesis) by
Vl+ 1(w) =lim e¢(zlrJ+1P(z)Vt(w). Z--->0
(2.2.33)
Note that we label vertex operators by their ¢-charge.
2.3 D-Branes and Orientifolds
2.3.1 R-R Charges
As described in the previous section, it is consistent for open strings to have their
ends fixed on a hypersurface. However, for different theories the surfaces may only
have certain dimensions. A T -duality transformation, whereby a dimension is com-
2.3. D-Branes and Orientifolds 24
pactified on a circle whose radius is taken to zero, is a legitimate transformation
of the theory. It is equivalent to reversing the sign of antiholomorphic modes on
that dimension; for closed strings, the modes winding round the circle exchange
roles with the propagating modes as the circle shrinks, generating an effective non
compact dimension. The effect on the R - R sector is to change the chirality; it
interpolates between I I A and I I B theories. The massless states in this sector are
tensor products of fermions, and hence bosonic. Thus if we write the holomorphic
and antiholomorphic basis spinors Sa and S~, the wavefunction of the total state is
uavf3 BaS~, which can be decomposed as
9
uavf3 S S' = '""'C S (CTI11 ···11n)a{3 S' a {3 ~ Ill ····lin a {3 (2.3.1) n=O
where f11 1 ···lin is antisymmetrised on its indices, and 6 is the charge conjugation
operator. The functions Cn _ C111 , ... 11n are antisymmetric forms (with C0 a scalar).
The equations of motion for them can be derived from constraints on the spectrum.
Since the basis spinors have definite chirality, we find that different dimensions of
forms are allowed for the different type-I I string theories:
IIA:
JIB: (2.3.2)
T-duality changes between the above forms. We expect them to couple to objects in
the effective action which carry R- R charge, and indeed it can be shown that they
are D-branes and orientifold planes. D-branes are the hypersurfaces to which open
strings may fix their end-points. For open strings, which have no winding modes,
the T-duality transformation exchanges Neumann and Dirichlet boundary conditions
and thus changes the dimensions of any Dirichlet hypersurfaces -· D-branes- present.
The terms in the effective action induced by the branes are proportional to
(2.3.3)
where the volume of integration is the (p+ 1 )-dimensional volume of the Dp-brane, p
specifying the number of spatial dimensions. The coupling of the forms to Dp-branes
2.3. D-Branes and Orientifolds 25
tells us that only certain dimensions of branes are allowed:
IIA: p = 0,2,4,6,8
JIB: p= -1,1,3,5,7,9 (2.3.4)
The D( -1)-brane is a special case, with even Dirichlet boundary conditions on the
time direction, and this corresponds to an instanton. In fact, however, if we choose
Dirichlet boundary conditions along the time direction for a brane with Neumann
conditions along p + 1 spatial directions, we have an extended instantonic object
denoted an Ep-brane; we shall discuss these further in section 2.4.5 and chapter 4.
Note that the term in the effective action is a tadpole, which for compact dimen
sions will give rise to infinities in scattering amplitudes if it is not cancelled. This
can be done using anti-D-branes, but to preserve supersymmetry we require Ori
entifold planes. Orientifold planes are a special object which carries negative R-R
charge, and are defined as a hypersurface invariant under a reflection and worldsheet
orientation reversal:
(2.3.5)
where Det(R) = -1. Orientifold planes arise from T-duality of unoriented strings,
and their properties can be deduced in that way.
2.3.2 Gauge Theory: Chan-Paton Factors
For open strings with both ends attached to a D-brane, the massless states in the
NS sector form a vector 'lj!i12 j0), 1-l = O .. p, which is interpreted as a gauge boson.
As remarked in the previous section, its wavefunction has gauge degrees of freedom.
For just one brane it is a U(1) field. If, however, we introduce a stack of branes
occupying the same hypersurface, we can attach the ends to different branes, and
supply the additional degrees of freedom necessary; for N coincident branes, we have
a U(N) gauge boson. We can write the vertex operator with a Chan-Paton Factor
>., an N x N matrix representing the end-point states:
(2.3.6)
If we consider amplitudes involving strings with these factors, along each bound
ary we must have continuous Chan-Paton index, and thus the amplitude will include
2.4. Intersecting Brane Worlds 26
a factor of tr(>. 1 ... >.n) for n vertex operators on a given boundary. Hence amplitudes
are invariant under >. -----> U >.Ut for unitary U, and so we have operators in the adjoint
representation.
When we have orientifolds (and orbifolds) we may include a projection on the
Chan-Paton matrices in addition to the worldsheet and spacetime generators. These
are manifest through a matrix /o for an operator 0, and we must then insert them
into amplitudes; a boundary is considered closed up to projection by /o, and thus
the trace in the amplitudes may include them: tr(r0 >.1 ... >. n).
Note finally that we must also include Chan-Paton factors for the gauginos,
and states stretched between stacks of branes, which will have the different ends
transforming under different gauge groups.
2.4 Intersecting Brane Worlds
2.4.1 Introduction
Intersecting Brane Worlds are a class of toy models in type IIA string theory which
have proven popular due to their ability to reproduce many features of the standard
model while being simple compared to many other constructions. Recent reviews
are given in [3,4), and some examples of model building are given in [5-14]. They
consist of D6 branes wrapping IR4 x 'II' 2 x 'II' 2 x 'II' 2 in type IIA string theory, where the
properties of the low energy theory are determined by the wrapping cycles of the
branes on the compact six-dimensional manifold; as in other formulations of string
theory, the parameters of the field theory have geometrical interpretations, but in
this case they are particularly simple and many quantities can be calculated exactly.
In these models, each 'II' 2 is a manifold parametrised by two quantities: the
complex and Kahler structures. In complex coordinates they are Riemann surfaces
identified under X"'"" X"'+-!21riR]:e-io:K, X"'"" X"'+-!21riR~, where RJ: and R~ are
radii and o:"' is a tilt angle, which is restricted in orientifold models to the discrete
values 1r /2 ("until ted") or ~~ coso:"' = 1/2 ("tilted"). In terms of these quantities 2
2.4. Intersecting Brane Worlds 27
we can write the complex structure of the tori:
(2.4.1)
and Kahler modulus
T"' = iR~ R~ sin a"' = iT!;. (2.4.2)
In general there are restrictions placed upon the complex structure by the require
ment of supersymmetry, but the Kahler structure is unrestricted and must be sta
bilised to eliminated the associated massless fields: however, we shall not treat this
here, although it is a topic of ongoing interest in the literature.
The identifications above are associated with homology cycles [Ai] and [Bi] re
spectively. We can then write the homology cycle of brane a as
(2.4.3)
Then n~ and m~ are the number of times the brane wraps cycles closed under the
first and second of those identifications respectively, and together with a position
coordinate contain all of the information about the wrapping of the brane. The
volume of the brane can be conveniently written as La = IT!=l L~, with
L"' = 21ryf(n"' R"-) 2 + (m"' R"') 2 + 2n"'m"' R"' R"' cos a"' a al a2 aal2 · (2.4.4)
However, it is often more convenient to work in terms of the canonical [Ai] and [Bi]
basis that represent the paths
(2.4.5)
then have
R1 · [Ai] + [Bi] R
2 cos al
[Bi]· (2.4.6)
We can then define ih~ = m~ + n~ ~~ cos ai to write the three-cycle wrapped by a
D6-brane a as 3
ITa= IJ(n~[Ai] + m~[Bi]). (2.4.7) i=l
2.4. Intersecting Brane Worlds 28
2.4.2 Mode Expansions and Quantisation
It is simple to quantise the string in such intersecting brane worlds, and thus obtain
the vertex operators. We refer the reader to [15, 16]. We first consider two non
compact complex dimensions X 1 and X 2 with two branes meeting at an angle n(Jl,
and the infinite strip worldsheet with coordinate w, 0::; R(w) ::; 1/2, -oo < <S(w) <
oo. The string mode expansion is just a free wave with different Dirichlet boundary
conditions at the ends. Using complex dimensions X= ~(X1 + iX2 ), we can write
the mode expansion
oX(w) = Lak-oe-irrw(-k+0-1),
k
;:)X- ( ) _ ""- -irrw(-k-0-1) u w - Lak+oe ,
k
We then combine the fields using the "doubling trick" , giving a theory on the infinite
cylinder:
{ oX(w) R(w) > 0
oX(w) = _ _ - , -oX(w) R(w) < o
(2.4.9)
and similarly for oX(w). The boundary conditions thus become a quasi-periodicity
on the cylinder:
oX(w- 1) = e 2rri08X(w)
oX(w- 1) = e-2rrw o.X(w) (2.4.10)
which is identical to the conditions for the holomorphic sector of closed strings on
an orbifold - except with a continuous angle. This correspondence allows much
of the technology to be adapted from orbifold calculations to intersecting brane
calculations. Since the correspondence is to half the amplitude, it may appear
that the calculations are simpler - however, this turns out not to be the case, as
complications arise due to the non-rationality of the angle, the brane wrapping in
compact spaces, and for loop diagrams the amplitudes are no longer modular forms.
1Throughout, in common with the literature, we shall use labels corresponding to angles with
1r factored out, and often refer to the labt:;l as an angle.
2.4. Intersecting Brane Worlds 29
The above boundary conditions and mode expansions are valid for an anti
clockwise rotation angle (} from the brane on the boundary R( w) = 0 to that at
~(w) = 1/2. However, we could have chosen a clockwise rotation, in which case we
would replace(}~ 1- e. A similar substitution will occur if we swap the branes at
each boundary. We shall persist with our choice in this section, but the reader should
note that in appendix (A.3) in the interests of generality no choice is specified.
Having quantised the string with these boundary conditions, we find that the
vacuum (which is annihilated by all positive frequency modes) does not correspond
to the identity operator in the state-operator mapping. Indeed, in sowing the above
worldsheet into a scattering amplitude, the emission or absorption of a string with
ends on different branes necessarily changes the boundary conditions on the scat
tering worldsheet: we need to insert a boundary changing operator a9 (w). We
determine its OPEs to be
8X(z)ao(w) rv (z- w)9- 1ro(w)
8X(z)ao(w) rv (z- w)-9r~(w) (2.4.11)
where r( w) and r' ( w) are "excited twist fields". The conformal weight is hue
~e(1- e).
Having discussed the bosonic contribution, for the superstring we must consider
the worldsheet fermions. Worldsheet supersymmetry determines that the periodicity
of the fermions on the infinite cylinder must be complementary to the bosons, so for
(post-doubling-trick) complex fermions we have
w(w- 1) = e21rive-Z1ri9w(w)
W(w- 1) = e-21rive21ri9W(w). (2.4.12)
Thus the mode expansions are simply shifted by an amount e. To obtain vertex
operators for the fermions, we can bosonise as usual, giving w( w) "" eiH(w) and
W(w) rv e-iH(w), and the requirement that the vacuum is annihilated by positive
frequency modes determines the vacuum operator to be
Av = ei(9+v-l/2)H(w) (2.4.13)
2.4. Intersecting Brane Worlds 30
with conformal weight ~(0 + v- 1/2)2. Notice that the degeneracy of the Ramond
sector is broken: spinors will only have components in the parallel dimensions.
2.4.3 Phenomenology
We may now consider the spectrum of strings in intersecting brane worlds: we
include three compact complex dimensions and consider intersections in each. For a
given pair ofbranes, we label the intersection angles in each dimension 0,_, /'1, = 1, 2, 3.
Including the contributions from ghosts, we find that the Ramond ground state
is still massless, but the energy of the Neveu-Schwarz ground state is determined
by the intersection angles. Thus, they determine the amount of supersymmetry
preserved by the branes. To preserve some supersymmetry they must be related
by an SU(3) rotation- (in the same way that the space group of an orbifold must
be a discrete subgroup of SU(3) to preserve some supersymmetry) - for parallel
branes, there will be an N = 4 supermultiplet, while for one angle zero and the
other two complementary there will be an N = 2 hypermultiplet, and finally for
1r(01 + 02 + 03 ) = n1r there will be an N = 1 chiral multiplet. These conditions are
satisfied provided that we preserve supersymmetry in the closed string sector: the D
branes must wrap special lagrangian (sLag) cycles of the manifold. This is achieved
by ensuring that the angles made by a brane relative to the orientifold plane sum to
zero mod 27f. While we would ultimately like a theory with broken supersymmetry,
directly breaking by adjusting the angles involves, for most models, fine tuning,
and moreover will generically lead to NS sector tadpoles signalling that the true
vacuum is actually a supersymmetric state. Hence we require a different method
of supersymmetry breaking, or at least a way to stabilise the branes at their non
supersymmetric angles; we shall thus consider only supersymmetric configurations.
As previously mentioned, each stack of coincident branes has an associated gauge
group dependent on the number of branes in the stack. This leads to gauge groups
of the form U(N). However, since U(N) = SU(N) x U(1), the U(1) factors are
anomalous, and acquire a string-scale mass, except for certain linear combinations
(see for example [17]). The nett result is that the hypercharge is a composite of the
U(1) fields from several stacks of branes.
2.4. Intersecting Brane Worlds 31
Family replication occurs naturally in these setups: for a given non-parallel pair
of branes on a compact manifold, they will generically intersect more than once as
they both wrap around their homology cycles. For the toroidal background, we can
write the number of intersections between two branes a and b as 3 3
lab= II I:b =II (n~m~- m~n~) (2.4.14)
This then determines the number of generations for a given pair of gauge groups.
The Higgs fields are localised at the intersection between a stack of branes cor
responding to the SU(2) and one (or several) corresponding to the U(1)y. The
potential can be realised by giving the scalar a tachyonic mass [18, 19], by adjusting
the intersection angles (the configuration will then condense by brane recombination,
corresponding to the Higgs acquiring a vacuum expectation value). The tree-level
Yukawa couplings for these models are determined then by the interaction of this
field and fields at other intersections. The amplitude is dominated by worldsheet
instantons - i.e. the classical action corresponding to stretching the worldsheet be
tween the three intersections as a conformal map, which corresponds to the area
of the triangle formed between them (and all possible wrappings around the tori).
These were calculated in [15, 19, 20], and the results are also summarised in appendix
A.l. Thus these models provide a simple geometric interpretation for the Yukawa
couplings, which aids model building; the coupling is proportional to e-Area/21rr:x',
where Area is the area of the triangle formed in each sub-torus between the inter
sections of the branes associated to the three fields coupled.
It would superficially appear that such a framework easily explains the Yukawa
coupling structure: consider three stacks of branes a, b and c, with three branes in
stack a, two in b and one in c. Then the H u field is localised at the intersection
between band c; Qi is between a and b; and (Un)c is between a and c (a is a CP index
counting the generation, and the superscript c denotes the conjugate field). We then
have lab = lac = 3, he = 1 to provide the correct number of generations, and the
superpotential term Yaf3Q£Hu(U~)c is suppressed by the exponential of the areas of
triangles between these. However, as established in [19], since we have a "left brane"
and a "right brane", we have a product form of the Yukawa coupling: due to model
building constraints, we have I~b = 3, I~b = 1, I;:b = 1 and I~c = 1, I~c = 3, I~ = 1,
2.4. Intersecting Brane Worlds 32
for k -I- l -I- m, and thus the exponential factors arise in different tori, giving
(2.4.15)
for vectors Aa, Bf3. This matrix has rank one, and therefore only allows one non
zero eigenvalue; this is a problem for model-building in toroidal intersecting brane
models, and we shall present a possible solution thereof in chapter 4.
2.4.4 Tadpole Cancellation
As remarked in the previous section, the R-R charges of the branes must be can
celled for the theory to be consistent (there are other constraints, but R-R charge
cancellation is the most relevant for this thesis). The condition for this is a simple
geometric one: writing J.-t6 for the charge of a D6-brane; K2 for the 10-dimensional
gravitational coupling; Na for the number of branes a in a stack on homology cycle
ITa on the compact dimensions; and a' denoting the images of brane a under any
orbifold, the relevant terms in the low energy effective action are
The equations of motion are
(2.4.17)
Since the left hand side of the above is exact, it integrates to zero, and the tadpoles
are thus cancelled if the overall three-cycle wrapped by the branes and orientifold
planes is trivial in homology:
(2.4.18) a
It is worth noting that homology does not completely describe the R-R charges
of the branes; since the branes carry gauge fields on them, there are additional K
theory constraints. For the cases of interest, these are not as restrictive as the above
equations, and we refer the reader to [21-23].
2.4. Intersecting Brane Worlds 33
2.4.5 Instanton Calculus
This subsection reviews some of the framework outlined in [24] and the references
therein, the most relevant of which are [25, 26].
The non-renormalisation theorem ensures that for N = 1 supersymmetric in
tersecting brane worlds, the low energy field theory will have a perturbative su
perpotential that receives no corrections beyond tree level. However, there may be
nonperturbatively generated terms which receive corrections up to one loop. In the
field theory these are produced by instantons; in string theory, instantons are man
ifest in the form of the Euclidean branes mentioned earlier, which have Dirichlet
boundary conditions along the time direction. For finite energy instantons (that
we expect should contribute to physical processes) any Neumann directions must
be a subset of those in the compact space; the Ep brane appears as a point in
four-dimensional Minkowski space. The way that the Ep brane is embedded in the
compact manifold then determines the properties of the instanton.
It is expected that Ep branes contribute to type IIA and liB models; however,
for the specific case of type IIA intersecting D6 brane models with six compact
dimensions, we can only have contributions from EO, E2 and E4 instantons, and
of these the most relevant are the E2 branes, since the toroidal manifold has no
continuous one or five cycles. It is possible that E4 instantons with fluxes are
relevant, but we shall not consider them in this work. Moreover, we shall only
consider E2 instantons wrapping special lagrangian cycles of the compact manifold,
as we expect only these to contribute to the superpotential
Strings attached to D-branes can be interpreted as fluctuations of the branes, and
massless strings as moduli. For example, gauge multiplets (with both ends attached
to the same brane) may be interpreted as moduli parametrising the deformation
and translation of the brane. For E2 branes, the attached strings cannot carry
momentum, and so it is not meaningful to discuss massless modes. However, the
moduli of the instanton consist of modes which satisfy the physical state condition,
equation (2.2.19). These may either have both ends attached to the E2 brane, or
an end on the E2 brane and one on a D6 brane of the model. These are known as
charged and uncharged zero modes respectively (reflecting their charge under the
2.4. Intersecting Brane Worlds 34
gauge groups). In string computations, only the charged zero mode vertex operators
may carry a factor of the string coupling (equal to .Jfh, g8 being the closed string
coupling) while the uncharged zero modes and chiral superfields may not.
In this work only E2 brane configurations for which there are no bosonic zero
modes are relevant; they would cause the amplitude to vanish in the cases of interest
to us, but in cases that they do not the formalism is treated in [27), which appeared
after the work of chapter 4 and [28]. For the fermionic zero modes, there are always
uncharged present; the presence of charged zero modes depends on whether the E2
brane intersects some of the D6 branes of the model. For left-handed uncharged
fermionic zero modes, there are two types: ,\~ and ;\.~, determined by the sign of the
intersection number IE2,a with positive for,\~. This determines the angle present in
the vertex operator (whether CJ¢ or a 1_¢)· The superscript index labels the charge
under the gauge group: there is a different zero mode for each brane in the stack a.
The superpotential is holomorphic in the chiral superfields, and thus if we con
struct a string amplitude holomorphic in the chiral superfields, we will be guaranteed
to calculate a contribution to only the superpotential in the effective lagrangian. If
we perform such a calculation, it will in general depend upon the zero modes of
the instanton, and so for our physical superpotential we must integrate these out.
In a field theory calculation, the integration over zero modes is accompanied by a
measure determined from the field configurations; in a string theory calculation, the
measure actually appears when we include all possible contributing string diagrams.
When a euclidean brane is present, we must include in an amplitude every possible
worldsheet that contains a boundary on the E2 brane: we must sum an infinite
number of disconnected worldsheets.
Another consequence of holomorphy is that the superpotential may only depend
upon the string coupling through a term of the form O(e-8n
2/ 9b ), where 9E2 is an
effective coupling of order #s· Since disk worldsheets have an attached factor of
g;l, then the origin of this in the light of the above discussion is clear: it arises
from including the sum of products of any number of disk diagrams with no vertex
operators inserted, divided by a combinatoric factor. Moreover, since one loop par
tition functions contain no coupling factor, we must include every such worldsheet
2.5. Non-Commutative Geometry 35
with one boundary on the E2 brane (when both boundaries lie on the E2 brane, the
contribution vanishes by supersymmetry). However, we must also explicitly remove
the zero mode contribution to the partition function, since they are to be integrated
separately (or equivalently, we could state that they have factors of the string cou
pling present that render them inadmissible). We shall elucidate the calculation of
these in section 4.2.2, but note that the total contribution to the measure is
exp [z'M(E2 06) + "'""Z'A(D6 E2) + z'A(D6' E2)] = Pfaff'(DF) (2.4.19) ' ~ a, a' vfdet'(Dn)
where the right hand side is the field theory interpretation in terms of pfaffians
and determinants, the primes indicate omission of the zero modes, and we have an
orientifold so must include image branes D6~ and the Mobius contribution between
the E2 brane and its image given by Z'M (E2, 06).
Finally, then, denoting uncharged zero modes by ()i, the contribution to a super
potential term Tin <I>n will be given by
J 4 II II II j II -k 2 2 Pfaff'(DF) II Wnp :J d X d()i dAa dAa exp( -87f I g E2) I ( <I>n). i a j k V det (DB) n
(2.4.20)
The superfields in (Tin <I>n) can be split between disk and annulus worldsheets; every
possible configuration must be included. For the disks, to ensure no further 9s
dependence, two charged fermionic zero modes must be included. Additionally,
clearly every zero mode in the model must be saturated by one in (Tin <I>n), since
they are Grassmann variables. Arranging these such that they are non-zero then
proves to be quite restrictive, and will be examined further for certain models in
chapter 4.
2.5 Non-Commutative Geometry
So far we have considered the background BJ.Lv = 0. However, if we now give the
antisymmetric tensor a constant value, we still have a consistent background for
string theory - but some interesting consequences. The matter action in conformal
gauge becomes
(2.5.1)
2.5. Non-Commutative Geometry 36
where Sm is the action (2.1. 7). We also make the substitution 77J.Lv ~ gJ.Lv in Sm,
where gJ.Lv still contains constant components but may involve a change of scale. In
the presence of a worldsheet with boundaries (i.e. we consider open strings on a
brane) this changes the boundary conditions to
(2.5.2)
where On and Ot are derivatives normal and tangential to a boundary respectively,
and the above are understood to be evaluated only at boundaries; the differentials
dz and dz are along the boundary. Note that it is also possible to add boundary
terms to the action (giving a vacuum expectation value to the gauge fields on the
brane), the effect of which is to modify the field strength B ---+ F = B- dA.
The above action results in noncommutativity between the target space coordi
nates (and non-anticommutativity between the fermionic coordinates). To see this,
if we compute the correlator of two target space fields on the disk (i.e. compute the
Greens' Function) mapped to the upper half plane by insisting that it respects the
(quantum) equation of motion and boundary conditions we obtain [29]
(XfL(z)Xv(w)) = -a' [gJl.V log I z- ~I+ QJl.V log lz- wl 2 + OJ.LV I log~- w]' z- w 2na z- w
(2.5.3)
where, writing g 9J.Lv' B = BJ.Lv'
QfLV 1 g 1 g + 2na'B 1- 2na'B
OJ.Lv -(2na'? 1
B 1
, (2.5.4) g + 2na'B g- 2na'B
and GJ.LV = c-1. Restricting the fields to the boundary <::J(z) = 0, we obtain
where
c(~) = { -1 x < 0 .
1 X> 0
(2.5.5)
(2.5.6)
If we then reduce to a field theory by taking the limit a' ---+ 0 with a transformation
of the metric:
g rv E (2.5.7)
2.5. Non-Commutative Geometry 37
then the logarithmic term vanishes from the above correlator, but we obtain a field
theory where the commutator of the spacetime coordinates is given by
(2.5.8)
This allows us to obtain a noncommutative field theory from string theory. Note
that in the above limit, gravity is decoupled. If, instead of taking the limit c -> 0
we merely consider it small (i.e. consider the effect of a high string scale, consistent
with observation), then there will be an effect on gravity. In chapter 5 we shall
investigat·e some of the consequences of this.
Chapter 3
Intersecting Brane Worlds at One
Loop
This chapter investigates several aspects of calculations at one loop involving states
at D6-brane intersections. The initial technology for such calculations was presented
in [30], but here we refine and extend that to new types of string diagrams. The
motivation is the discovery of a divergence in the two-point diagram; we consider
the consequences of this. We also consider some consequences for phenomenology of
the running of the Yukawa couplings in such models. This work is published in [31].
3.1 Introduction
As outlined in section 2.4, the perturbative superpotential for toroidal intersecting
brane models is well known, since it results from a disk calculation. However, the
effective field theory of the models is also specified by the Kahler potential, which
is renormalised (in principle) to all orders. Currently the Kahler potential in chiral
matter superfields is known to only quadratic terms and only at tree level [32, 33] (one
loop corrections to the moduli sectors of Kahler potentials in liB models have been
calculated in [34]). For a multitude of phenomenological reasons it is something we
would like to understand better, especially its quantum corrections. In this chapter
we go a step further in this direction with an analysis of interactions of chiral matter
fields at one-loop.
38
3.1. Introduction 39
@ Q u
Figure 3.1: An annulus contribution to 3 point functions .
Figure 3.1 shows the physical principle of calculating a one-loop annulus for
the example of a 3 point coupling, discussed in ref. [35). Take a string stretched
between two branes as shown and keep one end (B) fixed on a particular brane,
whilst the opposing end (A) sweeps out a triangle (or an N-sided polygon for N
point functions). Chiral states are deposited at each vertex as the endpoint A
switches from one brane to the next. As the branes are at angles and hence the
open string endpoints free to move in different directions, the states have "twisted"
boundary conditions, reflected in the vertex operators by the inclusion of so-called
twist operators. Working out the CFT of these objects is usually the most arduous
part of calculations on intersecting branes. The corresponding worldsheet diagram
is then the annulus with 3 (N) vertex operators. In the presence of orientifolds
there will be Mobius strip diagrams as well. There is no constraint on the relative
positioning of the B brane (although the usual rule that the action goes. as the
square of the brane separation will continue to be obeyed), and it may be one of the
other branes. (The case that brane B is at angles to the three other branes was not
previously possible to calculate: we shall treat it in detail in this chapter.)
Finding the Kahler potential means extracting two point functions , but because
the theory is defined on-shell we have to take an indirect approach; complete answers
can be obtained only by factorising down from at least 4 point functions. To do
this we first develop the general formalism for N-point functions in full. However it
would be extremely tedious if we had to factorise down the full amplitude including
the classical instanton action for every N-point diagram. Fortunately the OPE rules
for the chiral states at intersections allow a short cut: first we use the fact that the
superpotential is protected by the non-renormalisation theorem which has a stringy
incarnation derived explicitly in [30). In supersymmetric theories, the interesting
3.2. One-Loop Scalar Propagator 40
diagrams are therefore the field renormalisation diagrams which we can get in the
field theory limit where two vertices come together. A consistent procedure therefore
is to use the OPE rules to factor pairs of external states onto a single state times
the appropriate tree-level Yukawa coupling. In this way one extracts the off-shell
two-point function.
This procedure allows us to consider various aspects of one-loop processes. For
example, in N = 1 supergravity the non-renormalisation theorem does not protect
the Kahler potential, and only particular forms of Kahler potential (essentially those
with logdet(Kij) = 0 where Kij is the Kahler metric) do not have quadratic diver
gences at one-loop order and higher (for a recent discussion see [36]). It is natural to
wonder therefore how string theory ensures that such divergences are absent. Here
we show that the conditions for cancellation of the divergences is identical to the
Ramond-Ramond tadpole cancellation condition (essentially because in performing
the calculation one factors down onto the twisted partition function). (At the level
of the effective field theory there are two well known forms of tree-level Kahler metric
that are consistent with the one-loop cancellation of divergences, the "Heisenberg"
logarithmic form, and the "canonical" quadratic form). As a second application, we
consider the subsequent determination of field renormalisation; we find agreement
with the power law running that one deduces from the effective field theory. How
ever we also see that as we approach the string scale the power law running dies
away as the string theory tames the UV divergences.
3.2 One-Loop Scalar Propagator
3.2.1 Four-Point Amplitudes
String theory is defined only on-shell, so in order to calculate the energy dependence
of physical couplings we must calculate physical diagrams and probe their behaviour
as the external momenta are varied. To obtain the running of Yukawa couplings in
intersecting brane models, we should in principle consider the full four-point ampli
tudes, since they are the simplest diagrams with non-trivial Mandelstam variables.
However here we shall take the more efficient approach outlined in the introduc-
3.2. One-Loop Scalar Propagator 41
tion: due to the existence of a consistent off-shell extension of these amplitudes, it
is possible to calculate three-point diagrams to obtain the Yukawa couplings. This
was done in ref. [30] for the amplitude in certain limits (albeit with some inevitable
ambiguities). There it was shown that for supersymmetric amplitudes the non
renormalisation theorem held as expected, and the low-energy behaviour is entirely
dominated by the renormalisation of propagators. The latter can be consistently
calculated in full by factorising down from the full four-point functions, and this is
what we shall do here.
We shall focus on four-fermion amplitudes, since they factorise onto the lower
vertex amplitudes of interest - the Yukawa renormalisation amplitude and the two
scalar amplitude. In intersecting brane models, the allowed amplitudes are con
strained by the necessity of the total boundary rotation being an integer; super
symmetry then dictates the allowed chirality of the vertex operators via the GSO
projection. In the case of N = 1 supersymmetry, the requirement of integer rota
tion results in two pairs of opposite chirality fermions being required. In the case
of N = 2 or 4 supersymmetry, we are also allowed to have four fermions of the
same chirality. The correlator for the non-compact dimensions in the case of N = 1
supersymmetry was calculated in ref. [37], in order to legitimise their calculations
on orbifolds. We have checked that the N = 2-relevant amplitude gives the same
limit upon factorisation. Thus, we can simply use their result to justify using the
OPE behaviour of the vertex operators in order to obtain the low-energy limit of
the four-point function, and we can proceed with the calculation of the scalar prop
agator to obtain the corrections to the Yukawa couplings. In the next subsection we
outline some of the technology required, and in the ensuing subsections we extract
the information about the running of the couplings.
3.2.2 Vertex Operators
To begin, we review the technology, assembled in ref. [30], for the calculation of loop
amplitudes involving massless chiral superfields. The appropriate vertex operator
to use for incoming states at an intersection depends on the angles in each sub
torus. There are several possible conditions for supersymmetry (we will focus on
3.2. One-Loop Scalar Propagator 42
N=1 supersymmetric models) of which we will use the most straightforward: for
intersection angles ()"- (where K, runs over the complex compact dimensions 1 to 3)
we have 2..:,_ ()"- = 1 or 2. We have two possible conditions because each intersection
supports one chiral and one antichiral superfield, with complimentary angles. With
this condition the GSO projection correlates the chirality of the fermions with that
of the rotation, and we obtain vertex operators as in [15]:
(3.2.1)
3
V (ab) ( k ()"" ) , (ab)e-~uaS-r.eik·X IT ei(0"-1/2)H"' ,...0(~b) (z) -1/2 u, ' 'z = /\ ~ v ..
for 2..:,_ ()"- = 1, and for the "antiparticle":
\ (ab) -¢ ik·X IT -i{}"' H" (ab) ( ) /\ e e e IJ 1_ 0,. z (3.2.2)
3
V (ab) ( k ()"- ) , (ab) _p_ f3S ik·X IT -i(0"-1/2)H" (ab) ( ) _ 112 u, ,1- ,z = /\ e 2u f3e e IJ1_ 0,. z ~>-=1
The worldsheet fermions have been written in bosonised form, where the coefficient
a in eiaH shall be referred to as the "H-charge". The spacetime Weyl spinor fields
are the left-handed Sa = e±HHo-7-l!) and right handed s(3 = e±1(7-lo+7-li). The oper
ators IJ~ab) are boundary-changing operators (here between branes a and b), whose
properties are discussed in Appendix A.l.
_x(ab) is the appropriate Chan-Paton factor for the vertex. We shall not require the
specific properties of these, but in the amplitudes we consider they are accompanied
by model-dependent matrices 'Yf which encode the orientifold projections. These
matrices have been described for many models (e.g. [5, 6, 38], but we will only need
the results given in [39] for 'llN or 'llN x 'llM orientifolds:
(3.2.3)
where {)k is a 'llN twist, wk is also present in 'llN x 'llM, and PnR{Jk = ±1. k(l) runs
from 0 to N- 1 (M- 1) for {)k (w1), where {)o = w0 = 1. In the last expression we
3.2. One-Loop Scalar Propagator 43
have used the notation EJk,l = {)k(;}, and SO for example in the Z2 X Z2 orientifold we
have PnRel,o = PnReo,l = PnRel,l = -1 and PnReo,o = 1. The massless four-fermion
amplitude that we shall consider is given by
(3.2.4)
where brane all is parallel to brane a (or ·a itself in the simplest case).
3.2.3 Field Theory Behaviour
The behaviour of the amplitude in the field-theory limit is found by considering the
momenta to be small. When we do this we find that, due to the OPEs of the vertex
operators, the amplitude is dominated by poles where the operators are contracted
together to leave a scalar propagator. From the previous section and the results of
Appendix A.1, we find that two fermion vertices factorise as
(3.2.5)
for L~~: VK = L~~: _AK = 1, and c~~~~}-v"-N' are the OPE coefficients, given in equation
(A.l.l9). The calculation involves a factorisation of the tree-level four-point function
on first the gauge exchange and then the Higgs exchange, and a comparison with
the field theory result [15].
Note that for consistency the classical instanton contribution should be included
in the OPE coefficients as well as the quantum part. This means that as we go on
to consider higher order loop diagrams the tree-level Yukawa couplings (including
classical contributions) should appear in the relevant field theory limits. In the
factorisation limit these two fermions yield the required pole of order one around z2 ;
we will obtain a similar pole for the other two fermion vertices. If we were to then
integrate the amplitude over z3 we would obtain a propagator 2c/~2 .k3 preceding a
three-point amplitude, and performing the integration over z4 , say, we would obtain
3.2. One-Loop Scalar Propagator 44
' ' \ I I I I
Figure 3.2: Feynman diagrams in the field theory equivalent to our limit; we have
factorised onto the scalar propagator, and consider only gauge and self-couplings.
another propagator 2a, ~1 .k4 ( = 2a, ~2 .k3 ), and have reduced the amplitude to a two
scalar amplitude.
In this manner the four-point function can be factorised onto the two point
function and reduces to
plus permutations. The factors y(cab) and y(ballc) are defined in the Appendix
(A.l.20) - they are the Yukawa couplings, derived entirely from tree-level corre
lators in the desired basis. G0bcccb is the inverse of the Kahler metric Gcbcccb for the
chiral fields Cbc in the chosen basis; note that we do not require its specific form,
which was calculated in [32, 33]. Thus we have reproduced the two Yukawa vertices
in the field-theory diagrams (figure 3.2), coupled with a propagator which contains
the interesting information about the running of the coupling. Note that if we had
taken the limit z1 -------+ z2 , z4 -------+ z3 then we would factorise onto a gauge propagator.
In the field theory the tree-level equivalent would have magnitude
(3.2.7)
while the one-loop diagram yields
(3.2.8)
where we put u = 2k1 ·k4 as the usual Mandelstam variable, and II(u) is the one-loop
scalar propagator which we have yet to calculate (N.B. IT(u) "'GcbcccJ. Thus if we
3.2. One-Loop Scalar Propagator 45
want the renormalised Yukawa couplings in some basis (for example the basis where
the fields are canonically normalised), we simply set all =a and write
(3.2.9)
where "permutations" accounts for the equivalent factors coming from the renormal
isations of the fermion legs. Alternatively (and more precisely) since the superpo
tential receives no loop corrections, the above can be considered as-a renormalisation
of the Kahler potential
(3.2.10)
3.2.4 The Scalar Propagator
The object that remains to be calculated is of course the scalar propagator II(u)
itself, which we can get from the following one-loop amplitude:
(3.2.11)
which represents wave-function renormalisation of the scalars in the theory since as
we have seen four-point chiral fermion amplitudes will always factorise onto scalar
two-point functions in the field-theory limit. We fix both vertices to the same
boundary along the imaginary axis, and one of these we can choose to be at zero
by conformal gauge-fixing. For the present, we shall specialise to the following
amplitude
(3.2.12)
where the world sheet geometry is an annulus, and where one string end is always
fixed to brane all, and the other is for some portion of the cycle on brane a. In
the latter region the propagating open string has untwisted boundary conditions so
that in these diagrams the loop contains intermediate gauge bosons/ gauginos. It is
therefore these diagrams that will give Kaluza-Klein (KK) mode contributions to
the beta functions and power law running. The alternative (where the states never
have untwisted boundary conditions) corresponds to only chiral matter fields in the
loops, is harder to calculate and will be treated later and in Appendix C.
3.2. One-Loop Scalar Propagator 46
We have allowed the fixed end of the string to be on a brane parallel to brane
a (rather than just a itself), separated by a perpendicular distance y"' in each sub
torus. Diagrams where y"' =/=- 0 correspond to heavy stretched modes propagating
in the loop and would in any case be extremely suppressed, but for the sake of
generality we will retain y"'.
The diagram we are concentrating on here is present for any intersecting brane
model, but in general IT(k2) receives contributions from other diagrams as well. For
orientifolds, the full expression is
IT(k2
) =LA~(~~+ L ( M~~~kek,la + M;~kek.lb) (3.2.13) c k,l
where M~~~kek,la is a Mobius strip contribution, which we shall discuss later. Using
the techniques described in Appendix A.2, we find the amplitude
(3.2.14)
where
(3.2.15)
where the notation is as follows. The Ov = { 1, -1, 1, -1} are the usual coefficients
for the spin-structure sum. The Bv are the standard Jacobi theta functions (see
Appendix A.4) with modular parameter it as on the worldsheet (so that we denote
Bv(z) - Bv(z, it) as usual). DA and D'S are the lengths of one-cycles, determined in
the Appendix to be
(3.2.16)
where L~ is the wrapping cycle length of brane a in sub-torus ""' and T.f is the
Kahler modulus for the sub-torus, given by R! R2 sino:"', R! and R2 are the radii of
3.2. One-Loop Scalar Propagator 47
the torus, a/\: is the tilting parameter, and generally
Nv are normalisation factors that we will determine in the next subsection. Finally
the classical (instanton) sum depends on two functions Ll\: and Tl\:
1
Ll\:(q, 01\:) = 1: dzw1 (z) 2
(3.2.17)
where w1(z)= ( e1(z) )0"-
1e1(z-el\:q)
e1(z- q) e1(z- q) (3.2.18)
and where the contour for Ll\: is understood to pass under the branch cut between 0
and q. Note that this contrasts with the four independent functions that we might
expect from the equivalent closed string orbifold calculation.
3.2.5 Factorisation on Partition Function
The normalisation factors for the amplitude Nv must still be determined by fac
torising on the partition function (i.e. bringing the remaining two vertices together
to eliminate the branch cuts entirely) and using the OPE coefficients of the various
CFTs. We should obtain
3
A(ab) rv q-1-L:,. 0"2a'k2g2tr('"Va)tr(>.(ab) >,(ba))Z (it) II c<aba)O 2(a) o 1 aa 0",1-0" (3.2.19) 1\:=1
. for 2..:1\: el\: :::; 1, where 9o is the open string coupling, c~~~~~OO" is the OPE coefficient
of the boundary-changing operators determined in Appendix A.1 and Zaa(it) is the
partition function for brane a. It is given by
4 3
Zaa(it) = (87r2a't)-2'T](it)- 12 L bvO~(O) II z: (3.2.20) v=1 1\:=1
where
(3.2.21)
3.3. Divergences 48
is the bosonic sum over winding and Kaluza-Klein modes. The factorisation occurs
for q _____. 0, when LK. _____. 1 and TK. _____. it, giving
(3.2.22)
3.3 Divergences
With our normalised amplitude, we are finally able to probe its behaviour. The
important limits are q -'--> 0 and q _____. it where there are poles; in the former case the
pole is cancelled because of the underlying N = 4 structure of the gauge sector, but
in the latter it is not, and it dominates the amplitude. We should comment here
about a subtlety with these calculations which does not apply for many other string
amplitudes: due to the branch cuts on the worldsheet, the amplitude is not periodic
on q _____. q + it. This causes the usual prescription of averaging over the positions of
fixed vertices to give a symmetric expression to break down (this was an ambiguity
in [30]): the gauge-fixing procedure asserts that one vertex must be fixed, which,
to remain invariant as t changes, is placed at zero. The non-periodic nature of the
amplitude is entirely expected from the boundary conformal field theory perspective:
due to the existence of a non-trivial homological cycle on the worldsheet, we have
two OPEs for the boundary changing operators, depending on how (i.e. whether)
we combine the operators to eliminate one boundary. We have already used the
expected behaviour in the limit q _____. 0 to normalise the amplitude, but q _____. it
yields new information, namely the partition function for string stretched between
different branes: in this limit we obtain
3
A (ab) ,....., (it - q)-1-I: .. 0"'-2a'k22o/ k2g2tr(rva)tr(,\ (ab) A (ba))Z (it) II c(bab)O 2(a) o 1 ab 0"',1-0"'' (3.3.1) K-=1
where Zab(it) is the partition function for string stretched between branes a and b,
given by
(3.3.2)
3.3. Divergences 49
For supersymmetric configurations, however, this partition function is zero, and
hence we were required to calculate our correlator to find the behaviour in this limit.
It is straightforward to show that our calculation gives this behaviour: as q -t it,
the functions LK. diverge logarithmically, and so we perform a Poisson resummation
over n'B -which reduces the instanton sums to unity. Simple complex analysis gives
(3.3.3)
and we also require the identity [39]:
(3.3.4)
so that, for the N = 1 supersymmetric choice of angles we obtain (after Riemann
summation):
A~(:~ ( q, t) '""' (it- q)-l-2a' k
22k2e<I>tr(!'a)tr(.\ (ab) A (ba)) ~: 2rr(2rrvfal)3GcabC\a (8rr2a't)-2
(3.3.5)
The above is clearly singular in the limit k2 - 0, where the integral over q is
dominated by the behaviour at q = it, since at q = 0 the effective N = 4 SUSY of
the gauge bosons cancels the pole. Using the usual prescription for these integrals
we obtain the exact result for the amplitude, and find it is divergent:
dqA(ab) (q t) = e G - tr(>.(ab) >,(ba))~ dtt-2. 1u <I> NI 1oo 0 2(a) ' 4( a')3/2 CabCba Lb 0
(3.3.6)
This divergence is effectively due to RR-charge exchange between the branes,
and so we look to contributions from other diagrams (as given in (3.2.13)) to cancel
it. These consist of other annulus diagrams where one end resides on a brane other
than "a" or "b", and Mobius strip diagrams.
Fortunately we do not need to calculate the amplitude of these additional con
tributions in their entirety; indeed we can obtain all that we need from knowledge
of the partition functions and the properties of our boundary changing operators,
along with a straightforward conjecture about the behaviour of the amplitude. For
an annulus diagram where one string end remains on a brane "c", while the other
end contains the vertex operators and thus is attached to brane "a" or "b", we obtain
3.3. Divergences 50
two contributions: one from each limit. From the OPEs of the boundary-changing
operators, we expect to obtain
A~(~~ (x, t) = 2a' g5k2tr( ,·{)tr(,\ (ab) ,\ (ba)) (x )-2-2a'k2 ( c<aba) Zac + c<bab) Zbc) (3.3. 7)
where x now denotes q or it- q in the appropriate limits, and C(aba) is understood to
be the product of the OPE coefficients for each dimension. Considering the partition
functions (3.3.2) and the behaviour of the expression (3.3.5), then we propose that
the effect of the division by zero in the above is to cancel one factor of x with a
factor of 01(0). Hence, if we write equation (3.3.6) as AoNL,~ab, we thus obtain for
the divergence,
A(ab) _A (Ncfac Ncfbc) 2(c) - 0 La + Lb (3.3.8)
Note that cis allowed to be any brane in the theory, including images under reflection
and orbifold elements.
We must now also consider the contribution from Mobius strip diagrams, which,
with the same conjectured behaviour as above, give us
M (ab) 2 I 2k2t ( \ d ( f"/.R6k,la)* a ) -2-2a1
a,nRek,l = a 9o r "'U'l lnRek,l /nRek,l q
(c(aba)M + C(bab)M )
a,nRek,la b,nRek,lb (3.3.9)
where Ma,nRek,la denotes the Mobius diagram between brane a and its image under
the orientifold group ORE'Jk,la, supplemented by a twist insertion E'Jk.l to give a
twist-invariant amplitude, given by [39]
(3.3.10)
where 61'>, [5] is zero for a orthogonal to the 06k,1-plane in sub-torus l'l,, and 1 other
wise; dll is the number of sub-tori in which brane a lies on the 06k,z-plane, and no6k l
is the number of times the plane wraps the torus with cycle length L06 . Here OK, k,l
is the angle between brane a and the 06k,!-plane. We also define I:,a6k,t to be the
3.4. Running Yukawas up to the String Scale 51
total number of intersections of brane a with the 06-planes of the class [06k,z] in
sub-torus ""· For dil non-zero we have zero modes:
(3.3.11)
where f-L = 0 for untilted tori, and 1 for tilted tori. To obtain the divergence, the
same procedure as before can be applied once we have taken into account the relative
scaling of the modular parameter between the Mobius and annulus diagrams. Since
the divergence occurs in the ultraviolet and is due to closed string exchange, to
do this we transform to the closed string channel: we simply replace t by 1 I l for
the annulus, and 1 I ( 4l) for the Mobius strip. This results in an extra factor of 4
preceding the Mobius strip divergences relative to those of the annulus diagrams,
which are due to the charges of the 06-planes being 4 times those of the D6-branes.
We thus obtain the total divergence
A~= 4(:;312 GcabCba tr(_A(ab) _A(ba)) 100
dl{ ~a (2::: NJac- 4 L Prmf>k,tla,06k,l) 0 c,c' k,l
+ ~b (~ NJbc- 4 L Prmek,th,o6k,t) } (3.3.12) c,c k,l
Note that the total divergence is of the same form as that found in gauge coupling
renormalisation. We recognise the terms in brackets as the standard expression for
cancellation of anomalies, derived from the RR-tadpole cancellation conditions [6]:
(3.3.13)
where [ITa] is the homology cycle of a etc; note that the phases Prmf>k,t will be the
same as the sign of the homology cycles of the orientifold planes. Hence, we have
shown that cancellation of RR-charges implies that in the limit that k2 ---+ 0, the
total two-point amplitude is zero.
3.4 Running Yukawas up to the String Scale
Having demonstrated the mechanism for cancellation of divergences in the two point
function, we may now analyse its behaviour and expect to obtain finite results. As
3.4. Running Yukawas up to the String Scale 52
mentioned earlier, by examining the amplitude at small, rather than zero, k2 , we
obtain the running of the coupling as appearing in four-point and higher diagrams
where all Mandelstam variables are not necessarily zero. Unfortunately, we are
now faced with two problems: we have only calculated the whole amplitude for one
contribution; and an exact integration of the whole amplitude is not possible, due to
the complexity of the expressions and lack of poles. However, this does not prevent
us from extracting the field-theory behaviour and even the running near the string
scale, but necessarily involves some approximations.
Focusing on our amplitude (3.2.15), which in the field theory limit comprises the
scalar propagator with a self-coupling loop and a gauge-coupling loop, we wish to
extract the dependence of the entire amplitude upon the momentum. Schematically,
according to equation (3.2.9), we expect to obtain for Yukawa coupling Y, as k2 ---+ 0,
YR- Yo rv A+ B + 92132lnk2 + ,6. + O(k2)
87r (3.4.1)
where A represents the divergent term, the third term is the standard beta-function
running, and the fourth term comprises all of the threshold corrections. This is the
most interesting term: as discussed in [30] it contains power-law running terms, but
with our complete expressions here we are able to see how the running is softened
at the string scale. The term denoted B is a possible finite piece that is zero in
supersymmetric configurations but that might appear in non-supersymmetric ones:
in the field theory it would be proportional to the cutoff squared while in the (non
supersymmetric) string theory it would be finite. This term would give rise to the
hierarchy problem. Had we found such a term in a supersymmetric model it would
have been inconsistent with our expectations about the tree-level Kahler potential
- in principle, it could only appear if there were not total cancellation among the
divergent contributions.
The power-law running in the present case corresponds to both fermions and
Higgs fields being localised at intersections, but gauge fields having KK modes for
the three extra dimensions on the wrapped D6 branes [40]. For completeness we
write the expression for KK modes on brane a with three different KK thresholds
J-io,1,2 (one for each complex dimension - note that in principle we have different
3.4. Running Yukawas up to the String Scale 53
values for each brane a):
2 ( A 3
X [( )li l ( )i) 9a ~ ~ li !Lii P,i ~ = - 2 (;3 - 13) ln- + jJ L - -- - 1 IT -.- + ~s
81r P,o 8 !Lii-1 . , Mt-1 li=1 t<a
(3.4.2)
where {p,fJ} = {(L~+l)-I,Aj p,0 < p,1 < p,2 <A}, and A is the string cutoff which
should be 0 ( 1) for our calculation. {X li} = { 2, 1r, 47r /3} is the correction factor for
the sum, and ~s is the string-level correction. jJ and /3 are the beta-coefficients for
the standard logarithmic running and power-law running respectively. Note that the
above is found from an integral over the Schwinger parameter t' where the integrand
varies as (t')-612- 1 in each region; t' is equivalent to the string modular parameter
t, modulo a ( dimensionful) proportionality constant.
To extract the above behaviour, while eliminating the divergent term, without
calculating all the additional diagrams, we could impose a cutoff in our q integral
(as in [30]); however the physical meaning of such a cut-off is obscure in the present
case, so we shall not do that here. As in [30], we shall make the assumption that the
classical sums are well approximated by those of the partition function for the gauge
boson. However, we then subtract a term from the factors preceding the classical
sum, which reproduces the pole term with no sub-leading behaviour in k2 ; since the
region in q where the Poisson resummation is required is very small, this is a good
way to regulate the pole that we have in the integrand when we set k2 to zero. To
extract the ~ terms, we can set k2 to zero in the integrand: this is valid except for the
logarithmic running down to zero energy (i.e. large t), where the k2 factors in the -2o/k2
[~i~6~e--'f(<J(q)) 2 ] term regulate the remaining C 1 behaviour of the integrand
when the t integral is performed. This behaviour is then modified by powers of
(87r2a't/(L~) 2 ) 112 multiplying the classical sums after each Poisson-resummation, as
t crosses each cutoff threshold, yielding power-law running as expected; this was
obtained in [30], and so we shall not reproduce it here.
For A - 2 > t 1"-J 1, we have an intermediate stage where the KK modes are all
excited, but we have not yet excited the winding modes, represented by the sums
3.4. Running Yukawas up to the String Scale 54
over n'B. As a first approximation, the integral, with our regulator term, is
(3.4.3)
where we have used U;.(i>.., it)T"'(i>.., it) ~ it. If we had not set k2 to zero in the
integrand, the second term in brackets would be
·-2a'k2 ( _ >..)-l-2a'k2 t - 1 (
-2clk2
)
z t + 2a'k2t ·
The above can then be integrated numerically to give
A(ab) l"o.J e<I> k2
. G - (X! dtP(t) 2(a) Sn(a')2J2 CabCba Jo (3.4.4)
where P(t) is plotted in figure 3.3. It indicates how the threshold corrections are
changing with energy scale probed. The figure clearly demonstrates the softening of
the running near the string scale. Alas, we also find that the amplitude still diverges
(negatively) as t---+ 0, and so we conclude that the subtraction of the leading poles,
rather than fully including the remaining pieces (i.e. the other annulus diagrams
and the Mobius strip diagrams), was not enough to render a finite result. However
we believe that the condition in eq.(3.3.13) ensures cancellation of the remaining
divergences as well.
Despite this, the procedure of naive pole cancellation still enhances our under
standing of the running up to the string scale, because the divergent pieces (depend
ing as they do on much heavier modes) rapidly die away as t increases. This allows
us to focus on the behaviour of the KK contribution. The quenching of UV diver
gent KK contributions is well documented at tree-level but has not been discussed
in detail at one-loop. At tree-level the nett effect of the string theory is to introduce
a physical Gaussian cut-off in the infinite sum over modes. For example a tree level
s-channel exchange of gauge fields is typically proportional to [41, 42]
""'IT, e-f3"'M~"'a' ~ s-M2 ' !l !l
(3.4.5)
where {3"' = (27/1(1) - 7/J({)"') - 7/1(1 - ()"')) and where !vii = I:,, M~ .. is the mass
squared of the KK mode. Note that the above is similar to results in [43, 44]. The
3.4. Running Yukawas up to the String Scale 55
p
t
Figure 3.3: A linear plot of the "running" Yukawa coupling in modular parameter t ,
lower graph. The peak is very close tot= 1, i.e. the string scale. The top graph is
the standard power-law behaviour continued. The middle graph is the field theory
approximation using string improved propagators as defined in the text.
physical interpretation of this expression is that the D-branes have finite thickness of
order #. Consequently they are unable to excite modes with a shorter Compton
wavelength than this which translates into a cut-off on modes whose mass is greater
than the string mass. This suggests the possibility of "string improved propagators"
for the field theory; for example scalars of mass m would have propagators of the ·
form
(3.4.6)
where m2 = I.:,.(m''Y, and we are neglecting gauge invariance. To test this expres
sion we can follow through the field theory analysis in ref. [40] (which also neglects
gauge invariance): the power law running is modified:
(3.4. 7)
For comparison this curve is also included in figure 3.3. Clearly it provides a better
approximation to the string theory up to near the string scale, justifying our propa
gators. (Of course the usual logarithmic divergence in the field theory remains once
the KK modes are all quenched.) Also interesting is that this behaviour appears
in our first approximation; we expect it to appear due to T"(i>., it) and L"(i>., it)
3.5. Further Amplitudes 56
differing from the values we have assumed, where we would also have to take into
account the Poisson resummation required near ,\ -----+ t. We could thus attempt
an improved approximation by modifying these quantities: however, the integrals
rapidly become computationally intensive, and we shall not pursue this further here.
3.5 Further Amplitudes
Having discussed the one-loop scalar propagator and its derivation from N-point
correlators of boundary-changing operators, we may wish to consider calculating
the full N-point amplitudes themselves. The general procedure was outlined in [30]
where three-point functions were considered. The complete procedure for general
N-point functions is presented in Appendix A.2, the main new result being equation
(A.2.12).
Furthermore, we may also wish to calculate diagrams in which there are only
chiral matter fields in the loop; although we were able to use the OPEs to obtain the
necessary information pertaining to the divergences, we would need these diagrams
to study the full running of the scalar propagator, for example. We have constructed
a new procedure for calculating them, and describe it in full in Appendix A.3; the
crucial step is to modify the basis of cut differentials.
In calculating the contribution to the scalar propagator from diagrams with no
internal gauge bosons, we find some intriguing behaviour. In the field theory, the
diagrams which correspond to these (other than the tadpole) involve two Yukawa
vertices, and hence we should expect the result to be proportional to products of
Yukawa couplings; we expect to find the angular factors and the instanton sum.
We already see from the CFT analysis that the divergences are not proportional to
these which is perhaps not surprising. However when we calculate the finite piece
of diagrams, although we do obtain the expected instanton terms in the classical
action it is not immediately obvious that we obtain the Yukawa couplings here either.
Yukawas attached to external legs will always appear by factorisation thanks to the
OPE, but this is not the case for internal Yukawas. Hence, it seems likely that the
field theory behaviour of composition of amplitudes is only approximately exhibited
3.6. Conclusions 57
at low energy, and near the string scale this breaks down. This will be a source
of flavour changing in models that at tree-level have flavour diagonal couplings.
Unfortunately it cannot solve the "rank 1" problem of the simplest constructions
however, because canonically normalising the fields (i.e. making the Kahler metric
flavour diagonal) does not change the rank of the Yukawa couplings. We leave the
full analysis of this to future work.
3.6 Conclusions
We have presented the procedure for calculating N-point diagrams in intersecting
brane models at one-loop. We have shown that the cancellation of leading diver
gences in the scalar propagator for self and gauge couplings for supersymmetric
configurations is guaranteed by RR-tadpole cancellation. The one-loop correction
to the propagator is consistent with a canonical form of the Kahler potential in the
field theory, or one of the no--scale variety, where there are no divergences in the field
theory; had there been a constant term remaining, this would have corresponded
to a divergence in the field-theory proportional to the UV cut-off squared, which
would have been consistent with alternative forms of the Kahler potential. How
ever, other than the expected corrections from power-law running, we cannot make
specific assertions for corrections to the Kahler potential from the string theory.
When we investigated the energy dependence of the scalar propagator (in the off
shell extension, i.e. as relevant for the four-point and higher diagrams) we found that
there still remained divergences, which can only be cancelled by a full calculation of
all the diagrams in the theory. However information could be obtained about the
intermediate energy regime where KK modes are active and affecting the running.
We find the tree level behaviour whereby the string theory quenches the KK modes
remains in the one-loop diagrams, and we proposed a string improved propagator
that can take account of this in the field theory.
We also developed the new technology necessary for calculating annulus diagrams
without internal gauge bosons, and mention some interesting new features. At
present, however, the technology for calculating the Mobius strip diagrams does not
3.6. Conclusions 58
exist, and this is left for future work.
Chapter 4
Realistic Yukawa Couplings
through Instantons in Intersecting
Brane Worlds
This chapter studies some non-perturbative aspects of intersecting brane worlds,
and how they can resolve the "rank-one Yukawa problem" outlined in section 2.4.3.
This work is published in [28].
4.1 Introduction
Recall from chapter 2 that the perturbative superpotential in intersecting brane
models is given by a tree-level calculation (receiving no perturbative corrections) to
be of the form
and is thus of rank one, giving rise to only one massive generation. There have been
numerous subsequent attempts to solve this problem as well as other related analyses
of questions regarding flavour (e.g. refs. (35, 42, 45-51] , but many of these lost the
original link with geometry. In parallel there developed techniques for calculating
both tree level [15, 16, 20, 32, 52] , and loop [30, 31, 34] amplitudes involving chiral
(intersection) states on networks of intersecting D-branes.
The most recent development on the calculational side, whose consequences will
59
4.1. Introduction 60
be the subject of this chapter, has been the incorporation of the effect of instantons
in ref. [24] (for related work see also [53-55] . This work laid out in detail how
the contributions of so-called E2-instantons (i.e. branes with 3 Neumann boundary
conditions in the compact dimensions and Dirichlet boundary conditions everywhere
else) to the superpotential could be calculated. It also pointed out a number of ex
pected phenomenological consequences of these objects. Thus far attention has
mostly been paid to the fact that instantons do not necessarily respect the global
symmetries of the effective theory and so are able to generate terms that may oth
erwise be disallowed. In particular they can be charged under the parent anomalous
U(l)'s due to the Green-Schwarz anomaly cancellation mechanism. (Alternatively,
these charges can be associated with states at the intersection of the E2 and D6
branes.) For example this can induce Major ana mass terms for neutrinos which are
of the form
and which would otherwise be forbidden. In this equation gE2 is an effective coupling
strength which depends on the world-volume of the E2 instanton. These need not
be equal to the gauge couplings of the MSSM and the Majorana mass-terms can
be of the right order to generate the observed neutrino masses [24, 53]. Similarly
interesting contributions occur for the JL-term of the MSSM, yielding a possible
solution to the JL-problem.
In this chapter we reassess Yukawa couplings in the light of instanton contri
butions. In particular we claim that one-loop diagrams with E2 branes can solve
the rank-one problem and lead to a Yukawa structure which is hierarchical. All the
Yukawa couplings have by assumption the same charges under all symmetries, so
the extra terms are induced by E2 instantons which do not intersect the D6-branes.
The tree and one-loop contributions to Yukawa couplings are shown in figure 4.1;
the tree level diagram consists of the usual disc diagram with three vertices, the
one loop diagram is an annulus with the inside boundary being an E2 brane and
three vertices on the D6-brane boundary. By explicit computation of these one-loop
instanton contributions we show that the corrected Yukawas are of the form
y; _yo ynp et/3 - et/3 + et/3 '
4.1. Introduction 61
06 06
E2
Figure 4.1: Contributions to 3 point functions: tree level and one-loop instanton.
where the non-perturbative contribution has rank 3. Moreover as for the neutrino
Majorana masses, these contributions are exponentially suppressed by the instanton
volume. Hence the 1st and 2nd generation masses are hierarchically smaller than the
first.
In addition there is the possibility of making an interesting connection between
the Yukawa hierarchies and the Majorana neutrino masses. The suppression of the
latter with respect to the string scale should be similar to the suppression of 2nd
generation masses with respect to third generation ones. In general one sees that a
direct connection between compactification geometry and Yukawa couplings would
be manifest. The rest of this chapter is devoted to proving this result. In the next
section we outline the techniques of instanton calculus and compute the necessary
general results: these include the multiplying factors of disconnected tree-level and
one-loop diagrams with no vertex operators, which must appear in every such pro
cess. The section that follows provides the annulus correlator, and in particular
shows that the leading contribution yields instanton contributions to the Yukawa
couplings of the advertised form .
4.2. Instanton Calculus 62
4.2 Instanton Calculus
The general framework for calculating E2 instanton corrections to the superpotential
in string backgrounds was proposed in [24]. This section elucidates the technical
details of the calculus of E2 instanton corrections applied to toroidal orientifold
intersecting brane worlds.
4.2.1 Tree Level Contributions
The non-perturbative 0( e-81r
2 19'b) factor appearing in instanton contributions to the
superpotential is given by the product of all possible disk diagrams whose bound
ary lies along the instanton and with either no vertex operators or an RR-tadpole
operator. We shall label this contribution e-8~z. It is given by
so - 27r L E2- gs~ E2
(4.2.1)
where LE2 = rr:=l L£;2 is the volume of the instanton. The above can be given more
conveniently in terms of the gauge coupling on a reference brane a as
So _ S 2LE2 -2 E2- 1f La 9a
4.2.2 Pfaffians and Determinants
(4.2.2)
To perform any calculation in an E2 instanton background requires the knowledge
of the reduced Pfaffian and Determinant factor given by the exponential of the total
partition function of states intersecting the instanton with the zero modes removed.
In toroidal backgrounds, there are two classes of contributions to this: the first arises
when the instanton is parallel to a D6-brane in one torus, and the second appears
when there is no parallel direction.
One :parallel Direction
In this case, there is no zero mode associated provided that the branes do not
intersect, i.e. brane a is separated from the E2 brane by some distance Ya,E2 in a
torus i. In this case the partition function is given by
4.2. Instanton Calculus 63
Z(E2, D6a)
[ -81r
3o/t I i iT~si iya,E2Lb l2] ( 4 2 3)
x exp (Li )2 r + ' + 4 2 ' .. E2 a 7r a
After performing the sum over spin structures we obtain an expression that would
also be found in gauge threshold correction calculations
_ j k J dt ~ [-81r3o/t i iT_~si iYa,E2Lk2 2]
Z(E2, D6a)- NaiE2,aiE2,a t L_ exp (Li )2 lr +---;;;-- + 47r2a' I · r',s' E2
(4.2.4)
This expression can be integrated. When all Ya,E2 -+ 0 (where there are fermionic
zero modes that must be regulated) it gives [39]
_ i k [ (Lk2)2
Ti)l4 l Z(E2,D6a)y=O- -NaiE2aiE2a ln +lnlry(- +/'\; , ' ' a' a'
(4.2.5)
where tl; == '"'IE -log 47r is a moduli-independent constant. For non-zero Ya,E2, the
expression differs from those found previously; first we must Poisson-resum the
expression on r and s to obtain
Z(E2, D6a) = ( i )2
N Ji Jk L E2 _1_ (4 2 6) a E2,a E2,a T.i 81r2 · ·
2
x (XJ dl ~ exp [- l(Lb)2 (_!_m2 + ~n2) - 2in Ya,E2Lb]. Jo L 81r a' (Tl)2 21rTl 0 m,n 2 2
We then note the divergence for the piece m, n = 0 as l -+ oo (and note that
the separation between the branes will regulate any l -+ 0 infra-red divergence).
This arises in the NS sector and is cancelled for consistent models according to the
condition [39]
~ (Lk2)2N Ji Jk 4~ (Lb)2Ji Ik - 0 L Ti a E2,a E2,a- L Ti E2,06k E2,06k - ·
a={a,a'} 2 k 2
(4.2.7)
This is sufficient if there are no intersections between the E2-brane and the D6-
branes of the model; if there are branes for which there are no parallel dimensions
then the cancellation condition will include those.
We now rescale the variables and perform the integral:
exp [ - 2Jrin Ya,E2Lb] 1 t 271'2T.'
. k La 2 Z(E2, D6a) = Nal~2 a! E2 a- r· ( ') . , , 7r l I ·~ 12
m,n#O 2 m + z T~ n (4.2.8)
4.2. Instanton Calculus 64
This can be recognised as appearing in Kronecker's second limit formula:
(28'(z)Y 2::::: exp [27ri(mu+nv)]lm+nzl-28 = (m,n)~(O,O)
- 47rlog ie-7ru(v-uz)81(v ( ~z, z) I+ O(s- 1) (4.2.9) TJ z
and thus we obtain
· k [ 81 Cy~:;~b, ~) 2
_ Y~,E2 (Lk2)2] Z(E2,D6a)=-Naih,aiE2,a log (iT~) 27r3a' T~ ·
TJ a'
( 4.2.10)
This is a new result that also applies to gauge threshold corrections (as mentioned
earlier). It is noteworthy that the IR regularisation by separation of the E2 brane
from brane a does not commute with that used for Ya,E2 = 0. This reflects the
qualitative difference between E2 branes which wrap cycles where b1 (IIE2 ) = 0 and
those where the Betti number is non-zero; in the latter case (i.e. when the instanton
does not pass through a fixed point) it acquires additional uncharged fermionic zero
modes which must be integrated over, but when the separation between D6 and E2
branes reduces to zero there arise additional charged fermionic zero modes. We only
consider configurations with no additional zero modes in this chapter, and thus we
require the E2 branes to pass through fixed points and D6 branes to be separated
from these by some small distances. This is a common feature of many models,
although it is incompatible with the models of, for example, [14,56]. However, since
the extra uncharged modes carry no coupling, it is actually possible for E2 instantons
with six additional uncharged zero modes to contribute to the superpotential term
considered in this chapter generated at one loop, and thus we expect our conclusions
to extend to those models as well. The following analysis would be complicated by
the plethora of vertex operators, and thus we do not consider them.
No Parallel Directions
Here we have the raw expression
J dt ~ 8~(%) ry3 (it) fi3 Bv(¢£2,ait)
Z(E2, D6a) = Nale2,a - L 8v 2 (i!:.) n--( ) a (¢'" · ) · t v 81 2 Uv 0 f>=1 1 E2,a~t
(4.2.11)
However, this amplitude is not defined in the v = 1 sector, and additionally contains
fermionic zero modes which must be regularised. The first issue is straightforward to
4.2; Instanton Calculus 65
resolve: provided that the model satisfies RR-charge cancellation, the v = 1 sector
does not contribute. To address the second issue we must decide how to remove the
zero mode from the trace over states: this is trivial once we express the amplitude
in the open string channel. Since the integral over the modular parameter and
the exponentiation of the partition function translates the sum over states into a
product, it is clear that we must simply expand the partition function as a series in
e-1rt and remove the 0(1) term in the v = 1, 2 sectors. The piece subtracted turns
out to be cancelled over all the contributions by the RR cancellation condition, and
so we can ignore it.
We must now perform the sum over spin structures. To this end, we use the
identity
(4.2.12)
to write the partition function as
Z(E2, D6a) = NaiE2,a J ~t L Ov [e~(O)- Ov(0)82 log01(~)] v~l
1 rr3 Ov(¢>'E2,ait) 0
X ( ( ) (4.2.13) 2ne~ o) t~;=l e1 ¢lF;2,ait
It is straightforward to show that, if we were to include the v = 1 term to the above,
it would not contribute, and hence we can perform the spin structure sum using the
usual Riemann theta identities. The term proportional to Ov(O) vanishes, and we
have remaining an expression identical to that found in gauge threshold corrections.
The calculation is then that of [39], including the cancellation of divergences arising
in the NS sector. The result is then
4.2. Instanton Calculus 66
4.2.3 Vertex Operators and Zero Modes
Massless strings with at least one end on the E2 brane are the zero modes of the
instanton. The most important of these (in that they are always present) are the
fermionic modes with both ends on the E2 brane; these are the modes associated
with the two broken supersymmetries in the spacetime dimensions. The vertex
operator for toroidal models is thus
(4.2.15)
where the internal spin field is ea = [1~=1 e±~H"; it is the spectral flow operator
for the internal dimensions, or alternatively the internal part of the supercharge.
This generically has four components after the GSO projection, and thus we have
the eight modes of (broken) N = 4, but only the two modes that preserve the same
N = 1 supersymmetry as the branes and orientifolds in the model will contribute.
In practice this involves choosing the correct H-charges to transform a fermion into
a boson in the internal dimensions, and without loss of generality we shall take this
b [13 i H" to e "=1 e-2 .
Massless fermions at an intersection between the E2 brane and a D6 brane of
the model are internal zero modes of the instanton. They will not be relevant for
the following analysis, but are in general vitally important for calculations; we list
their vertex operators here for completeness:
~~;=1
3
v,-b = );i ,e-~ ~II ei(¢'fm-l/2)H" a " A b,/ </JE2b
(4.2.16) ~~;=1
where I is the intersection number running from 1, ... , [IIE2 nITa]+, and I' runs from
1, ... , [IIE2 n fib]- (note that <I>E2b- 1/2 = -(</>bE2- 1/2)); i = 1, .. , Na(Nb) is the
Chan-Paton index. .6. and ~ are the boundary-changing operators in the four non
compact dimensions which interpolate between Dirichlet and Neumann boundary
conditions; their OPE is
.6.(z)~(w) rv (z- w)-112 . (4.2.17)
4.3. Yukawa Coupling Corrections 67
4.3 Yukawa Coupling Corrections
A straightforward analysis of the possible diagrams in the instanton calculus shows
that in the presence of fermionic zero modes an E2-instanton cannot contribute
to a Yukawa term in the superpotential for the quarks. However, if there exists
one or more sLag cycles for which there are no intersections with any branes in
the model, then a contribution to this term is possible through a one-loop annulus
diagram. Typically this involves the separation of the E2-instanton from each D6
brane in one sub--torus only, so that were the E2 brane replaced by a D6 brane
wrapping the same three-cycle and the separations reduced to zero, this brane would
preserve different N = 2 supersymmetries with each of the branes in the model. This
possibility arises generically but not in every case in model-building, so implies a new
moderate constraint, in order to benefit from the consequences of these instantons.
The superpotential term generated by these instantons can thus be determined
from section 4.2. It involves three superfields on an annulus diagram, and thus two
fermions and one boson together with two fermionic supersymmetry zero modes.
We can then write
(4.3.1)
Since there are now three picture-changing operators in the above amplitude,
to obtain a non-zero result we must apply each operator to a different sub-torus
direction; this is since each internal fermionic correlator must have zero nett charge,
and thus the charges introduced by the supersymmetry zero modes must be cancelled
by those of the PCOs. The amplitude will then have no momentum prefactors, and
thus will not factorise onto a scalar propagator; this is explicitly shown in appendix
B.2.
Having determined the Pfaffian and Tree-level factors in section 4.2, we must
4.3. Yukawa Coupling Corrections 68
now determine the annulus correlator. The total amplitude can be written as
i=1
L · IT {f,(aXK,(y,.)a<t>~b(zi)a<t>'bJz2 )a<t>~Jz3 )) {Yl ,y2 ,y3 }=P(xl ,x2 ,X3) K-=1
X ( eiH"(y,.) ei(</>(ab) -1)H" (zl) ei(</>(bc) -1/2)H" (z2) ei(<f>(ca) -1/2)H"(z3) e- 4H"(w1) e- 4H"(w2))
(4.3.2)
where z1 has been fixed to 0, and the angles c/J~b' ¢'be and ¢~a are external (hence c/J~b + ¢'be +c/J'be = 2 and 2::!=1 ¢"' = 2). co.f3 = i(r1 r 2 )o:f3 is the charge conjugation operator.
The above amplitude can be evaluated using the techniques outlined in appendix A.2
(and [30, 31]). The perhaps unexpected form of the spin-field correlator is explained
in appendix B.l. The most non-trivial part of the above is that involving the
boundary-changing operators, which are dominated by worldsheet-instanton effects.
We split ax = aXqu +axel, for which aXqu has boundary conditions such that
all vertices have no displacements, whereas axel absorbs the displacements between
the vertices. We have N
(aXqu IT O"<J>J = 0 (4.3.3) i=l
and thus the amplitude is dominated by worldsheet instanton effects. To show the
above, we consider
(a.X(w) IT~1 a<t>;(zi))
(f1~1 a</>; (zi))
(8X(w) IT~ 1 a<t>;(zi))
(f1~1 O"<J>; (zi))
rv
( 4.3.4)
and construct a set of differentials satisfying the above local monodromies and pe
riodicity of the worldsheet; these are the differentials given in [30, 31, 57]. We then
apply the global monodromy
1 dwa.X + dw8X = iia "fa •
(4.3.5)
4.3. Yukawa Coupling Corrections 69
where Ia are a set of N paths on the worldsheet and Va are the displacements
between the vertices in the target space. There are N independent differentials that
comprise ax and ax' and so the global monodromies determine the coefficients by
linear algebra. Since the paths are independent, the equations are non-degenerate
and for va = 0 we must set all of the coefficients to zero, establishing the claim
above. Defining theN x N matrix W (i' runs through the set {z1 , z2 , ... , ZN-z} and
i" denotes the complementary set {zN_ 1, ZN} ):
1 dzwi' (z) "fa
1 dzii" (z) 'Ya
(4.3.6)
in terms of the N cut differentials { wi', wi"}, we can then write (after applying the
doubling trick to relate ax to ax) N N
(aX(x) II ac/>; (zi)) -va(W-1 )f,,wi" (x)(IT ac/>; (zi)) i=1 i=1
N N
(ax (x) II ac/>i (zi)) -va(W-1 )f,wi' ( -x)(II a¢; (zi)). (4.3.7) i=1 i=1
The correlator is thus directly proportional to the displacements. Specialising now
to the specific three-point function and using the prescription of [31] we have cycles
{ Ia} = {A, B, C2 } where A and B are the canonical cycles of the torus, and C2
is the path passing between two vertices on the worldsheet. For this diagram, in
each sub-torus there is one brane parallel to the E2 brane, and this represents a
periodic cycle on the worldsheet. The prescription for the amplitude requires that
we permute the vertices cyclically so that the periodic cycle passes along the real
axis; writing {a~~:, b~~:, c~~:} for the cyclic permutation of branes {a, b, c} such that brane
a~~: is parallel to the E2 brane in torus K, we have
(4.3.8)
where ~~~: is the shortest distance between the target-space intersections b~~:c~~: and
c~~:a~~: along the brane c~~:, and y~~: is the distance between a~~: and the E2-brane. The
4.3. Yukawa Coupling Corrections 70
phase ei<f>~K~K in the last line appears due to the orientation of brane a" relative to
c".
The exponential of the worldsheet instanton action depends upon the same dis
placements, and the amplitude can then be schematically written as
(Vlab V~2V~2Ve-l/2Ve-l/2 )a,E2 = c/J(ab)'I/J(bc)aCaf3'1/J(ca)f3((h(h- 8281)
J dt J dz,dz, t!l (Jr.n~ u::;;e-0) (4.3.9)
Note that the action which appears here is the one-loop action as derived from the
monodromy conditions, and depends on the integration variables. The crucial part
is that the functions fi arise from the different permutations of applications of the
picture-changing operators. Each choice of fi corresponds to a different contribution
that is separately factorisable across the tori (and different from the perturbative
Yukawa term owing to the v~ factors). Factorisability is in general lost upon per
forming the integral since there are no poles. However, since the functions are
different, even if the integrals were dominated by a particular region of the moduli
space, we would have Yukawa matrix corrections of the form
6
Yaf3 ~ 2:::: A~B~ . ( 4.3.10) i=l
This is a sum of six independent rank one matrices, giving a rank three Yukawa
matrix as advertised in the introduction to this chapter. Note that the correction
terms are suppressed relative to the perturbative superpotential by approximately
the factor e-5~2; this provides not only rank 3 couplings but an explanation for
the hierarchy in masses between the top quark and the others. At the level of this
analysis there is no obvious explanation for the hierarchies between the 1st and
2nd generation; this could yet arise from the non-factorisation of the worldsheet
instanton contribution. We leave this issue for future work.
Chapter 5
N oncommutativity from the string
perspective: modification of
gravity at a mm without mm sized
extra dimensions
In this chapter we focus on type liB string theory, using bosonic string theory
as a toy model, and consider the noncommutative field theory that arises when a
magnetic flux coupling to open strings on a brane is manifest in the macroscopic
dimensions. We examine the phenomenon of UV /IR mixing in this framework. We
then consider the effect on gravity, since the running of the closed string couplings is
affected by the open string theory, feeling the effect of the noncommutativity. The
work appears in [58].
5.1 Introduction
Gauge theories in which the coordinates are noncommuting,
(5.1.1)
are interesting candidates for particle physics, with curious properties (for general
reviews of noncommutative gauge theories see refs. [29, 59, 60]). One whose conse-
71
5.1. Introduction 72
quences we would like to understand a little better is ultra-violet(UV)/infra-red(IR)
mixing [61,62]. This is a phenomenon which gives rise to various pathologies in the
field theory, making it, at best, difficult to understand. In this chapter we set about
examining UV /IR mixing from the point of view of string theory with a background
antisymmetric tensor (BJLv) field, which provides a convenient UV (and hence IR)
completion. Along the way, as well as seeing how the pathological behaviour is
smoothed out, we will outline the characteristic phenomenology of this general class
of theories in the deep IR (i.e. at energy scales lower than those where noncommu
tative field theory is a good description): they resemble the B = 0 theories but with
Lorentz violating operators which can be taken parametrically and continuously to
zero by reducing the VEV of BJLv. As a bi-product we also show that the UV /IR
mixing phenomenon extends to the gravitational sector (although a field theoretical
interpretation for UV /IR mixing in gravity is difficult to obtain). This allows the
curious possibility that gravity may be non-Newtonian on much longer length scales
than those associated with the compact dimensions.
Because UV /IR mixing, and the particular problems and phenomena to which
it gives rise, are rather subtle, we begin now with a detailed discussion of exactly
what questions we would like the string theory to answer, after which we restate
our findings in more precise terms. UV /IR mixing has its origin in the fact that the
commutation relations intertwine large and small scales. At the simplest level, in
a gedanken experiment where x1 and x2 do not commute, the uncertainty relation
~Xl~X2 rv i()12 together with the usual Heisenberg uncertainty ~xl~Pl rv i imply
~x2 rv -{}12 ~p1 : short distances in the 1 direction are connected to small momenta
in the 2 direction and vice versa. At the field theory level, this intertwining of
UV and IR leads to the infamous phenomenon of UV /IR mixing in the non-planar
Feynman diagrams: nonplanar diagrams are regulated in the UV but diverge in the
IR. Essentially, contrary to the standard picture of the Wilsonian effective action,
heavy modes do not decouple in the IR so that, for example, trace U(1) factors of
the gauge group run to a free field theory in the IR even if there are no massless
excitations [63-67].
The agent responsible for these unusual and challenging features of noncommu-
5.1. Introduction 73
tative gauge field theories is the Moyal star product,
(5.1.2)
used in their definition. It induces a phase factor exp ~k.B.q in the vertices, where
k is an external momentum and q is a loop-momentum. This oscillating phase reg
ulates the nonplanar diagrams in the UV, which can most easily be expressed using
Schwinger integrals: for example the one-loop contribution to vacuum polarisation
takes the form ( c.f. [63-66, 68, 69])
J dt k2
IIJ.Lv(k) rv te- 4t .•• (5.1.3)
where kJ.L = ()J.Lv kv and the ellipsis stands for factors independent of k. The expo-
nential factor in the integrand is a regulator at t rv k2 rv k2 I Mf.c, where we define
the generic noncommutativity scale by (}J.LV = O(M!Vb). Thus the diagram, which
without this factor would be UV divergent, is regulated but only so long as k =/= 0.
The result is that the UV divergences of the planar diagrams reappear as IR poles
in k in the nonplanar diagrams.
These divergences are problematic. First they signal a discontinuity because the
k --+ 0 limit of the integrals is not uniformly convergent: physics in the limit () --+ 0
does not tend continuously to the commutative theory. Moreover they lead to alarm
ing violations of Lorentz invariance. For example, the lightcone is generally modified
to a lightwedge [68, 70]. This is in sharp disagreement with observation. Further
more in noncommutative gauge theory, the trace U(l) photon has a polarisation
tensor given by [62]
(5.1.4)
where the additional term rv II2 is multiplied by a Lorentz violating tensor structure.
It is absent in supersymmetric theories [62], but since supersymmetry is broken, we
expect it to be at least of order M~usY times by some factor logarithmic ink (where
Msusy is a measure of the supersymmetry breaking). The result is a mass of order
Msusy for certain polarisations of the trace-U(l) photon while other polarisations
remain massless [71]. Gymnastics are then required to prevent this trace U(l) photon
mixing with the physical photon.
5.1. Introduction 74
Clearly then, the outlook from the perspective of field theory is gloomy; because
the IR singularities are a reflection of the fact that field theory is UV divergent, any
attempt to resolve them without modifying the UV behaviour of the field theory is
doomed. With this understanding, the general expectation is for a more encouraging
picture in a theory with a UV completion, such as string theory. A more precise
argument is the following. First it is easy to appreciate that, without an explicit
UV completion, noncommutative field theory is unable to describe physics in the IR . .
limit. As noted in ref. [72] and in the specific context of string theory in ref [73, 74],
UV /IR mixing imposes aIR cut-off given by lkl >AIR= MAke. Inside the range uv
.. 1 AjR"' IBiiiAuv < lkl < Auv, (5.1.5)
the field theory behaves in a Wilsonian manner, in the sense that modes with masses
greater than the UV cut-off do not (up to small corrections) affect the physics there.
However outside this range the Wilsonian approach breaks down because modes
above Auv affect the physics below AIR. Indeed this inequality makes it impossible
to make statements about either the Oii ---+ 0 limit or the k ---+ 0 limit within field
theory. In other words, a UV completion is needed not only to describe physics
above Auv but also physics below AIR, and in particular to discuss the existence or
otherwise of discontinuities there. The picture is most obvious in the context of
running of gauge couplings. Between AIR and Auv the effective action accurately
describes the running of the trace U(1) gauge coupling regardless of what happens
above Auv. Below A1 R, UV physics intervenes. For example a period of power law
running due to KK thresholds in the UV is mirrored by the "inverse" power law
running in the IR. Now, the precise UV completion may take various forms, but A2
suppose for example that it acts like a simple exponential cut-off, e-~, in the
Schwinger integral. The planar diagrams are regulated in the usual manner, but
the nett effect of the noncommutativity for the nonplanar diagrams is that the UV
cut-off A~v is replaced by A;11 = 1/(k2 +A[!~) [61]. In this case when k « A[!~ (i.e. when we are below the IR cut-off) we would have
(5.1.6)
and normal Wilsonian behaviour would be restored, with the couplings matching
5.1. Introduction 75
those at the UV cut-off scale. Of course there is no reason to suppose that such a cut
off in any way resembles what actually happens in string theory, and to discuss the
nature of the theory below AIR requires full knowledge of the real UV completion.
What then are our general expectations for physics below AIR? Does it correspond
to an effective field theory? If so, what happens to the Lorentz violating divergences
in theIR?
These are the precise questions we would like to explore, using a framework in . .
which the noncommutative gauge theory is realised as a low-energy effective theory
on D-branes [29, 75-77]. Our arguments are based on the two point function as
calculated on D-branes in the background of a non-zero B-field [73, 74, 78, 79]. In such
a theory, taking the zero slope limit in a particular way [29] (a' --+ 0 with g11v rv a'2 )
yields a noncommutative field theory in which the role of the noncommutativity
parameter is played by the gauge invariant Born-Infeld field strength: indeed in this
limit the open string metric and the noncommutativity parameter are given by [29]
GJ1V - ( 1 1 ) JlV --g--g-F g+F
with F11v = 2na' B11v, B11v being the (magnetic) field strength, and
()JlV = -2na' (-1_p_1_)J1V g-F g+F
(5.1.7)
(5.1.8)
respectively (we will henceforth restrict ourselves to noncommutativity in the space
directions which we will label ij). The theory at finite a' provides a convenient
UV completion of the noncommutative gauge theory. The UV "cut-off" acquires
a physical meaning: it is the scale above which the noncommutative field theory
description is invalid and string modes become accessible, and is of order
Auv = 1/H. (5.1.9)
The IR "cut-off" is accordingly given by
(5.1.10)
and, likewise, physics below this scale is best understood by performing a string
calculation. We will rather loosely continue referring to the scale AIR as the IR
5.2. General remarks on UV /IR mixing in string theory 76
cut-off although of course we are chiefly interested in exploring the effective theory
below it.
What we will show in this chapter is that the one-loop effective theory in the
k ~ 0 limit (including any threshold contributions) is the same as the commutative
() = 0 theory, and in particular there are no IR divergences. Below AIR, physics
differs from the () = 0 physics only by nonsingular residual effects that are calculable
in any specific model, and we will estimate their magnitude. In addition we point
out that the two point function of the' graviton also gets stringy contributions at
one-loop which can modify gravity right down to AIR: if for example MNc rv lTeV
and M 8 rv Mp1, then gravity is modified at a mm even when there are no large
extra dimensions. This is an effect equivalent to the one described for the gauge
theory however there is no simple effective field theory description and it is difficult
to understand in terms of "planar" and "nonplanar".
The rest of the chapter is organised as follows. In the next section, we will
discuss and determine the general form of UV /IR mixing in noncommutative field
theory which is embedded in string theory. In section 3 and 4, the mentioned
general characteristics of UV /IR mixing will be justified with explicit amplitude
calculations based on bosonic and superstring models. In section 5, we will analyse
how the graviton two point function is modified by noncommutativity. In section 6,
we will discuss how noncommutativity in string theory may lead to a modification in
the IR property of gravity. We will also discuss its phenomenological implications.
5.2 General remarks on UV /IR mixing in string
theory
Assuming that the string theory amplitudes are finite (as issue to which we return in
due course), it is natural that the IR singularities should be cured in much the same
way as UV singularities are, since they are intimately related: they are essentially the
same singularities. It is also natural that string theory should cure discontinuities
afflicting the field theory; we certainly expect a string amplitude calculated at non
zero F, which is after all a rather mild background, to tend continuously to the
5.2. General remarks on UV /IR mixing in string theory 77
one calculated at F = 0. What is more striking is that in a non-supersymmetric
theory the Lorentz violating II2 term also tends to zero as k2 I a' below the IR cut-off,
reminiscent of the field theory behaviour with the naive Schwinger cut-off.
Consider nonplanar annulus amplitudes in bosonic string theory on a Dp-brane.
As we shall see, the general structure of a one loop diagram can be very heuristically
written as
A J dt - (p+l) _J.2 /4t !( ) NP"' -t 2 e t .
t . (5.2.1)
The function f(t) includes kinematic factors as well as sums over all the open string
states in the loop. The integration parameter t is the parameter describing the
annulus. In the field theory limit a' - 0 we recover the expected nonplanar field
theory contribution, with t playing the role of a Schwinger parameter. In addition
all but the massless open strings (and in this case the tachyon whose contribution
we discard) do not contribute in this limit. In the present discussion we are of course
not interested in taking the field theory limit but will instead keep a' finite. The
crucial feature of the amplitudes governing the IR behaviour is that the nonplanar
integrands always come with a factor e-1.2 /4t irrespective of whether we are above or
below AJR. When k2 » a' the integrand is killed everywhere in the stringy region
t < a' and the amplitude is close to the field theoretical result. Indeed one may
make a large t expansion rendering the amplitude identical to the field theoretical
one. On the other hand in the area of most interest below the IR cut-off we have
k2 « a' and hence stringy t < a' regions also contribute to the integral. If the
integrand is finite and free of singularities then in the limit as k - 0 the amplitudes
clearly tend continuously to their commutative equivalents. Thus the finiteness of
the string amplitudes immediately guarantees that the k - 0 limits and the () - 0
limits give the same physics. Moreover in this limit we may expand the e-1.2
/4t
factor inside the integral. The nett result is that far below the IR cut-off one-loop
amplitudes may be written as
p A(e, k) "'A(O, k)(l + ,\- + ... ),
a' (5.2.2)
where ,\ is a factor including loop suppression and gauge couplings and the second
piece is the leading term in the small k2 I a' expansion of the exponential factor. Note
5.2. General remarks on UV /IR mixing in string theory 78
that the A( 0, k) prefactor includes the usual one-loop contributions of the commu
tative theory and hence all stringy threshold corrections. Thus although various
compactification scenarios may result in vastly different threshold corrections, the
leading effect of non-zero B field will always be of this form. (Extension to N-point
amplitudes is trivial.)
Based on this generic expression for the amplitudes, phenomenology below A1 R
takes on a characteristic form. First from the low energy point of view the nett effect . .
of the non-zero B field is simply to take the non-planar contribution to thresholds
of gauge couplings and move them down to the IR cut-off, inserting between AIR
and Auv a region approximating conventional noncommutative field theory. Below
AIR, the leading deviation from the commutative theory (including all its stringy
thresholds) has a factor k2/a', with the dimensionality being made up by powers of
a'.
Thus for example the I12 term is of the form
(5.2.3)
in a non-supersymmetric theory and
I12(k2, P) "'>.(k2a')2M~~SY a
(5.2.4)
in a theory with supersymmetry softly-broken at a scale Msusy. (Note that the
factor of (k2o/)2 is simply to undo the power of Jc-4 in the above definition of rr2.)
This introduces a birefringence into the trace-U ( 1) photon, a polarisation dependent
velocity shift of order
(5.2.5)
This effect is much milder than the naive expectation and can be made phenomeno
logically acceptable with a large MNc even if the physical photon is predominantly
made of trace U(1) photon as described in ref. [72]. The model dependent issue
here which we will expand upon in the following sections is the coefficient >. which
encapsulates the strength of the one-loop contributions (i.e. threshold corrections
to couplings) relative to the tree level ones.
If the physical photon is decoupled from the trace U ( 1) photon (see for example
[80] where the trace U(1) photon becomes weakly coupled in theIR and forms part of
5.2. General remarks on UV /IR mixing in string theory 79
a hidden sector to break and mediate supersymmetry), then there can be interesting
implications for gravity. Consider a theory where the physically observed Planck
scale receives significant one-loop threshold corrections from the open string sector.
This contribution can be computed from the two point function of the gravitons
with the open string modes running in the loop. To gain some more intuition on
what effects of noncommutativity might be, we turn to an effective field theory
description. A reasonable (but, as it turns out, incorrect) guess for the effective . .
field theory coupling the open string modes to the graviton is a lagrangian of the
form
.c _ J d4 r--:; Jl.J.L' vv' FJl.V * FJl.'v' - Xy -gg g 4g2 ' (5.2.6)
where there is, note, no star product between the "closed string metric" or in its de-
terminant. The desired contribution can be computed from the two point function
of the gravitons with the gauge bosons running in the loop, and thus our effec
tive field theory above would generate "planar" and "non-planar" diagrams exactly
as in the pure gauge case, the crucial point being the presence of a Moyal phase
coming from the vertices. Thus one might expect that in string theory, turning on
a B-field would separate planar and non-planar contributions to the graviton two
point function, in much the same way as for the photon. Thanks to UV /IR mixing
the nonplanar contributions would change all the way down to AIR below which
they would asymptote to the values of the commutative theory. There, the leading
deviation in Planck's constant from that of the purely commutative theory should
be precisely as described above for the gauge couplings. As we will see in the sec
tion 5 the true picture is actually more subtle than this 1 . Nevertheless the effect
we described persists; namely that sub-leading k2 /a' suppressed corrections in the
two point function of the graviton lead to a modification of gravity at energy scales
higher than AIR·
1 In the field theory limit, an effective vertex involving a graviton and two photons exists (and
indeed we compute it), but there is no such simple Lagrangian from which it could be derived.
5.3. UV /IR mixing in the bosonic string 80
5.3 UV /IR mixing in the bosonic string
We will first look at the 2-point function on the annulus for pure QED, equivalent
to the noncommutative Yang-Mills action
S = -~~F *FJ.Lv 4 J.LV (5.3.1)
The contributions to the 2-point amplitudes on Dp-branes in a non-compact 26-
dimensional volume requires open string vertex operators
(5.3.2)
which have been appropriately normalised (g'b = (2rr)P-2gc(a')~). This gives the p
amplitude
roo (p+l)
-2a'gbp Vp Jo dt (8n2a't)--2-ry(it)-24 x (5.3.3)
r dx e-2a'k.G(x,x').k (cl.Gxx'·c2- 2a'(cl.Gx.k)(c2.Gx'·k))l · Jo x'=O
Here x, t play the role of dimensionless Feynman and Schwinger parameters respec
tively. At this point, we should comment that throughout this chapter we shall take
the fundamental domain of the annulus to be (0, 1/2] x (0, it].
Note that we write the measure with integration over all of the vertices, and then
use the annulus' translation invariance to fix one vertex, including a volume factor
oft. The one-loop Green's functions required depend on whether the diagram is
planar or nonplanar, and are given by (73, 78, 81]
(fJ2)af3 eaf3 Gaf3(x x') = l 8af3 + J + K-
' 0 12 I 1 a a (5.3.4)
where, for the planar case,
e (x-x' i) I p( ') l I 1 -t-, t I
0 X- X = og t ry3(ijt) '
and for the nonplanar case,
e (x-x1 i)
I NP( ') l 4 -t-, t o x - x = og t 7]3 ( i / t) '
KP(x- x') = -~E(x- x'), (5.3.5)
KNP(x + x') = ±~(x + x'), t
(5.3.6)
5.3. UV /IR mixing in the bosonic string 81
where the +(-) in KNP applies for the outer (inner) boundary. The feature of
these expressions which ensures the regularisation of the nonplanar diagram is the
contraction k.G.k appearing in the exponent of the integrand. We find
- 2a' k.GN P .k
-2a'k2 IP 0• -2
-2a'k2JNP- _k_. 0 81ra't
(5.3.7)
(5.3.8)
Having established the Green's functions, we can perform the integration by
parts in equation (5.3.4) to extract the kinematics. Defining (for ease of notation) A --4 rr2 = I12k ' we find
Af rrf(k2, o)[(EI. E2)k2 - (E1 · k)(E2 · k)J, (5.3.9)
A~p ITfP(k2, P)[(EI. E2)k2 - (EI. k)(E2. k)] + ft2(k2, P)[(EI. k)(E2. k)],
where the standard gauge running is given by the formula
rrl = 4a'2gbp 100 dt Z(t)e-~xa 1t dx e-2a'k2JoUo)2,
while the Lorentz-violating piece is given by
A 2 2 100 dt k2 1t 2 'k2JNP rr2 = 47r 9Dp 0 t2 Z(t)e-4.Tt 0 dx e- a 0 .
(5.3.10)
(5.3.11)
In eqn.(5.3.10) a = 0 or 1 for the planar or nonplanar case respectively. The term
Z(t) is the partition function for the model, which we shall take throughout to be
(5.3.12)
The parameter d is inserted to remove adjoint scalars from the theory: there are
p - 1 physical polarisations of the photon, and the remaining 25 - p modes are
scalars, so we can interpret the parameter d as removing d of these modes. This is
performed either by considering string theory in 26- d dimensions [78, 82], or taking
the spacetime to be, for example, IRp+l x IR25_p-d x IRd/ZN with the p-brane at a
singularity such that the orbifold projection removes the scalars [73]. In this way,
we can alternatively consider d as modelling the effect of compactified dimensions,
and hence we shall refer to d "compact dimensions" throughout.
The forms (5.3.9)-(5.3.11) for the 1-loop amplitudes were already discussed in
ref. [78] in the field theory limit. However it is important to note now that we
5.3. UV fiR mixing in the bosonic string 82
have not taken any field theory limit and yet the k dependence is already entirely
contained within the k2 term: the sole effect of noncommutativity is to truncate the
Schwinger integration to 2nt > :~, even in the full string expression.
Thus there are two regimes that we will consider. The first is the regime where
k2 I a' » 1. In this case the Schwinger integral is truncated to the region 2nt » 1
and the integral is well approximated by the t ---t oo limit. The second regime is
where k2 I a' « 1. In this case much of the Schwinger integral is over the region
where 2nt « 1 and one expects the t ---t oo limit to be a poor approximation. In
this limit a good approximation to the integral requires a modular transformation
of the '13 and TJ functions to the closed string channel. It is natural to think of a'
as playing the role of the UV cut-off to the field theory, a' · Aif~, and then this
regime corresponds precisely to
(5.3.13)
1.e. the region in the deep IR where the field theory computation breaks down.
Indeed the integral will become sensitive to the global structure of the compact
ified dimensions since the t ---t 0 UV end of it corresponds to closed string modes in
the deep IR. Note that [83, 84) studied the connection between IR poles and closed
string tachyons; we shall neglect these as we are interested in extracting results rele
vant to consistent theories. There may be other thresholds as well as the string scale
where for example winding modes of the compact dimensions start to contribute in
the integral. In order for there to be an effective field theory description below
AIR these effects should add contributions independent of p. In order to incorpo
rate these effects one can divide the Schwinger integral into regions t E [0, 1) and
t E [1, oo) where the two approximations are valid.
5.3.1 Brief review of planar diagrams
The methods for obtaining the low energy behaviour of string diagrams in order to
derive the effective four-dimensional field theories have been well covered elsewhere
[82, 85). Since the integrals do not contain any evidence of the non-commutativity,
as can be seen from the Green's functions, the only difference from the B = 0
5.3. UV fiR mixing in the bosonic string 83
calculation in this case is a phase dependence on the ordering of the vertices. The
reader is referred to the Appendix for details of the t ---+ oo limit for planar diagrams,
which reviews some of the basic techniques that we will be using. We shall consider
d dimensions transverse to the brane to be compactified with a radius close to or
at the string scale, although we shall pay special attention to the case d = 0. The
contributions to II1 and fh in this limit are given in equations (C.1.6) and (C.1.7).
The t ---+ 0 limit of the planar diagrams is the UV contribution, that is t E [0, 1]
indicating momenta much higher than the string scale. It is well known as the planar
contribution to the string threshold correction, however we present it here in order to
emphasise the way that string theory is thought to render such contributions finite.
Let us compute the t---+ 0 contribution to the two point functions for a Dp-brane in
26 non-compact dimensions. We modular transform the expressions to give for the
partition function
2 (p+l) 12 21< 2rr (81r a't)--2-t eT (1 + 24 e-T + ... ), (5.3.14)
where, since we are assuming no compact dimensions, there are no winding modes,
and thus
ITP ---+ 9bP 11 dt t (21;-p) (2t)-2a'k211 dy 1 uv (p-3)
161r2(87r2a')_2_ o o
[ 241 cos1ryl2l sin 1fYI_2_2a'k2 + 81 cos1ryl 2l sin 1fYI-2a'k
2]. (5.3.15)
Here we cannot neglect the "pole" pieces, but perform the integral in terms of beta
functions and analytically continue in the momentum, using
11 2 a+ 1 b + 1
dyl COS7rYial sin 1rylb = -B(--, -2-)
0 1f 2 (5.3.16)
to give the zero-momentum limit
(5.3.17)
a threshold contribution to the gauge couplings which is finite when p < 23.
Now, of course this computation is a cheat because it assumes that transverse
space was non-compact. In a compact space, sooner or later in the t ---+ 0 limit we
5.3. UV /IR mixing in the bosonic string 84
need to sum over winding sectors in the measure of the integral. Once the winding
sectors are included the effective p is p _ 25 and the integral diverges. However
this divergence is resolved in a way that is at least qualitatively well understood:
the natural way to write A2uv is with the parameter S = ~2
which reveals the
expression to be in the a'k2 «1 limit simply an IR pole due to a massless closed
string tadpole. Indeed in this limit and when p = 25 the contribution from level n
is proportional to
_4"" (n-1) dS -!;-(n-l)S 1a' dT 2 ' 1= 2
-2e r = e " ' 0 T a'
(5.3.18)
as appropriate for closed string states with m 2 = 4;/ (n- 1). Such tadpoles are a
signal that we are expanding around the wrong vacuum, and the solution is to give
a VEV to the relevant fields in order to remove them. In this way the background
is modified by the presence of the tadpoles and the nett effect is that systems with
> 2 codimensions (i.e. p < 23) are insensitive to the moduli of the transverse
dimensions, whereas those with 1 or 2 codimensions get threshold corrections that
are respectively linearly or logarithmically dependent on the size of the transverse
dimensions, but are still believed to be finite even when supersymmetry is broken by
the construction. In principle in certain tachyon-free non-supersymmetric cases one
can resum the tadpole contributions to the tree-level perturbation series to achieve
a finite result. The precise details are rather subtle and beyond the scope of this
work, and the reader is referred to refs. [86-88] for more details.
5.3.2 Non-planar diagrams in the k2 /a' >> 1 limit
We now turn to the non-planar diagrams. Once we turn on the B-field, the presence ;;,2
of the e- s""'t regulating factor will cause the celebrated UV /IR mixing. We may
treat the UV and IR contribution at the same time in the two limits k2 I a' « 1 and
k2 I a' » 1. Consider the second of these limits. The integrand is killed in the region
where t « k2 I a' and hence we may always use the large t limit of the integrand.
5.3. UV /IR mixing in the bosonic string 85
We obtain
(5.3.19)
where in the last step we assumed lkllkl « 1 or in other words momenta lkl « MNc·
In the case p = 3 this gives the same logarithmic running to a free field theory in
the IR observed in the field theory. When p > 3 we find power law running in the
IR as described in [89]. The Lorentz-violating term IT2 is given by
II2
= gb~ {oo dTT- cv~3) t dy (24- d)e-Tk 2 (y-y2)- !~
( 47r) 2 J 2-m' J 0
2
~ (24- d) gDv +I 2p-lr(p + 1)1kl-(p+l) (5.3.20) (47r)Ef'- 2
and shows a similar power law behaviour in the IR. For p _:_ 3 and d = 22 we
reproduce the result of [78]. This behaviour is entirely in line with what one would
expect from the field theory.
5.3.3 Non-planar diagrams in the k2/a' << 1 limit
In this limit one expects to find behaviour differing from noncommutative field
theory. We now have to split the integral into two halves, t > 1 and t < 1. The first
IR part is treated similarly to the previous section, except in this case we simply set
k = 0 in the integrand when we consider a' ----t 0, and should thus obtain the same
results as in the planar case; it is straightforward to show that for p > 3 2
d 9Dv 2 ( ')~ ~ 3(47r)E.:}!(p-3) 27ra 2'
2 9Dp 2 I -(p+I)
~ (24-d) E±!( )(21ra) 2.
(47r) 2 p + 1 (5.3.21)
The contributions are roughly constant, and equal to those of the k2 j a' » 1 limit
when k2 = 4a'.
5.3. UV fiR mixing in the bosonic string 86
The second, UV, contribution for t < 1 is the most interesting, as it is this
contribution which in field theory gives IR poles. We now modular-transform the 2rr
expressions, and expand in powers of e-T. For no compact dimensions, we have
2 1 11 gD 1 (21-p) 1 2 f<2 rro.'k2 rrftv -t P ~ dtf 2 (2f)-2Q k e-Brro.'te------u- dy Sin2 21ry.
16n2 (8n2a') 2 o o (5.3.22)
Note that for ~~ -t 0 the integration is finite and the integral goes continuously to
that of the commutative contribution, i.e. we have
(5.3.23)
as promised in the Introduction; in other words, at momenta k «AIR the Wilsonian
gauge couplings return to the values they would have had for a completely commu
tative theory with the same gauge group. Note that this statement is expected to
be true even when p ~ 23 and in compact spaces for the following reason. In the
finite examples we have seen, the effect of string theory is clearly to allow the limit
k -t 0 to be taken continuously, and to give the same physics as (} = 0. If this is true
of any consistent UV completion, then it seems reasonable to assume that what's
good for the planar diagrams is good for the nonplanar ones. In other words, if the
diagrams are formally divergent, continuity demands that the vacuum shifts which
remove the UV divergences (i.e. closed string tadpoles) in the B = 0 theory should
do so in the B =f. 0 theory as well, up to O(k2 fa') corrections. Note that this is true
even though the non-planar diagrams do not factorise onto disks in the closed string
channel; IR singularities arise from divergences in the partition function which are 1<2 rro.1k 2 .
regulated by the e- srrc>'t e------u- term, and so when these diVergences are cancelled,
so are the IR poles.
This reasoning leads one to expect that the Ih term is regulated, since it should
tend to zero as (} -t 0. Let us check this by computing the final contribution which
is
(5.3.24)
5.4. Supersymmetric Models 87
5.4 Supersymmetric Models
To include the effects of worldsheet fermions we require the fermionic propagators
[74]:
(5.4.1)
where the index v specifies the spin structure, which must be summed over in the full
amplitude. The above differs from the usual boundary fermion propagators purely
by the replacement of the metric by the open string metric, but when we perform
the rescaling of the external momenta and polarisations [78] it is transformed back
to the standard propagators:
which we shall use from now on.
--t bo:{3()v(zl - z2)B~ (0) Bv(O)()l (zl - z2)
bo:{3G~(z1 - z2), (5.4.2)
We wish to calculate the one-loop amplitude for two spacetime bosons with an
arbitrary amount of supersymmetry in the loop, which is defined by the compact
dimensions - and thus only affects the amplitude via the partition function. The
vertex operators are
(5.4.3)
and the resulting amplitude gives
and A 1CX) k2 1t / 2 Ih=41f2gbp dtL Zv(t)e-4a't dxe- 2o:klo,
0 . v 0 (5.4.5)
where Zv(t) is the partition function for the theory, and
ix-1/2. (5.4.6)
Thus the spacetime fermionic component does not contribute to the Lorentz-violating
term, since the kinematics for it are just the standard commutative gauge pieces.
5.5. The two point function of the graviton 88
The Lorentz-violating term thus derives from bosonic correlator exactly as in the
bosonic string, the only difference being the partition function. Of course, if there
is any supersymmetry then this term will vanish, as we expect, and the remain
ing Lorentz-preserving term can be calculated from the off-shell continuation of the
fermionic piece. For N;::::: 1 SUSY, II1 can be simplified using the identity
(5.4.7)
to give
II = 4g2 (cx')2100 dt"'""" e2o.'PJ z (t) e~ (0) 1t dxe-2o.' k2 Io 1 Dp L...-t l/ e (o) .
0 l/ l/ 0
(5.4.8)
Again this is essentially the usual expression for computing threshold corrections,
but with an exponential factor inserted for non-planar diagrams.
To summarise the results of this and the previous two sections, as advertised
in the Introduction, both bosonic and supersymmetric theories are found to tend
continuously to the B = 0 theory as k ---+ 0. In particular the couplings freeze out
below AIR and the entire region above A1R can now be consistently integrated out
in the usual Wilsonian manner. The phenomenological footprint of the non-zero
B field is then in the dispersion relation of massless particles, and in particular a
birefringence of the trace-U(l) photon, which gets a polarisation dependent velocity
shift of order ~ M§usyM'f
v"' c M4 NC
Whether the EM photon feels this effect is a model dependent question.
5.5 The two point function of the graviton
(5.4.9)
We now turn to the effect of the non-zero B field on gravity by focusing on the
graviton two-point function. In particular, consider the corrections to the Newtonian
force law of gravity due to the coupling of the graviton to gauge fields at one-loop.
The momentum dependence of these corrections determines the running of Planck's
constant, and our experience with gauge couplings suggests that this also may be
subject to UV /IR mixing.
5.5. The two point function of the graviton 89
In the naive extension of noncommutative field theory of eq. (5.2.6), the one
loop contributions divide into planar and non-planar exactly as they do for the
trace U(l) photon. However in string theory the relevant diagram is an annulus
with two graviton (closed string) vertices on the interior of the world sheet, and so
the only way that planar could be distinguished from nonplanar would be either for
there to be some kind of radial ordering effect in the vertices, or for there to be a
limit in which the major contribution to the diagrams came from when the vertices
were on the edges of the annulus. Neither of these possibilities is true and so, even
before making any computation, it seems unlikely that there will be a simple field
theory approximation involving Moyal products. The field theory limit has been
the subject of a recent study in ref. [90] where it was indeed found to be a rather
complicated issue. However for the present study we do not need to derive the
effective action (and indeed we don't): we will instead examine the modification
of the Newtonian force law between matter (open string) fields on the brane, by
looking at the two point function determined at the string theory level.
By restricting our attention to the force law between matter fields, we are evading
a significant technical difficulty, namely that in a sense we have two metrics, one for
open strings and one for closed. We wish to examine the momentum dependence of
the gravitational force between open strings confined to magnetised D-branes. In
principle we ought to be doing this by factorising a four point open string amplitude
on the graviton two point function. The relations for G, g and e imply
(5.5.1)
determining the coupling of the matter on the brane to gravity. We must choose a
coordinate system where the components of the metric gi-Lv are made small ( a/ FI-Lv)
for the dimensions in which magnetic field is turned on, so that the noncommutativ
ity tensor 01-Lv can be tuned to the desired values. Then the relevant momentum scale
for the amplitude is given by the Mandelstam variables running through the loop,
determined from the external momenta as contracted with the open string metric.
Importantly the closed string metric is vastly different in the regimes of interest:
for the exchange of a graviton with four-momentum qi-L between open strings, the
5.5. The two point function of the graviton 90
Mandelstam variables correspond to scales of order
(5.5.2)
but if we were interested in graviton exchange between external graviton states, it
would be more appropriate to use
(5.5.3)
which, by definition, for q rv AIR, would be of order rv M;. In the former case, as we
are only dealing with graviton propagators the difference is immaterial since we can
always rescale the graviton states to absorb the difference, but the correct procedure
for (or indeed physical meaning of) the latter is less clear. (For example we would
probably want more information about the other contributions in such a process
coming from B fields, and also more information about what the asymptotic states
are - i.e. the effective field theory.) Thus in calculating the graviton correlator, we
shall decompose the square of the momentum in terms of the open string quantities,
and consider k2 / ci » a' k2 , while k2 k2 « 1.
We shall restrict the discussion to exchanges between D3-brane states, for which
(since there can be no orientifold planes) we need only consider the annulus for the
induced gravity on the brane at one loop. A typical model for this scenario would
be D3/ D7-branes at a C6 /ZN orbifold singularity [33, 91], with a magnetic flux on
the 1 and 2 directions. However, we shall keep the discussion as general as possible.
We proceed initially as in ref. [92, 93] to extract the correction to ~b, denoted o, by considering the following kinematic portion of the graviton two-point function:
(5.5.4)
where the vertex operators are given by
Va(h, k) = (5.5.5)
g;Na ~,h11v(8XJ1.(z)- i~' : k · 'ljnj;Jl.(z):) (8Xv(z)- i~' : k. ;j;;j;v(z): )eik·X(z,z).
In the open string channel the Green's function is given by modular transforming
5.5. The two point function of the graviton 91
the result of ref. [74]2:
(5.5.6)
where
and where f(x) _ -[xjt], [y] denotes the closest integer to y. Thus the self-
contraction terms, with normal-ordering and the w 1 ---t w 2 limit performed, are
(5.5.8)
The fermionic Green's functions are obtained from the torus functions using the
doubling trick:
We obtain
(1/P(z)vi(w))v
({/:t( z){i( w) )v
(vP(z){i( w))v
go:flct(z- w),
go:flct(z- w),
(5.5.9)
( ( BGB)o:f3 ea!3 ) ,
-i go:f3 + 2 4n2(a')2 - 22na' Gt(z + w). (5.5.10)
2Note that this choice of propagator differs slightly from those given elsewhere [73, 78], but it
was asserted in [7 4] that the additional terms are necessary to ensure periodicity and obedience
to the equations of motion. They cause a discrepancy when the fields are taken to the boundary;
(5.3.5, 5.3.6) are not obtained from (5.5.7) . However, the closed string propagators only differ by
linear terms in Jc and Kc, plus the function f which plays no essential role in amplitudes (merely
ensuring that the derivatives of the logs in the antisymmetric portion contain no discontinuities).
The reader can check that (5.5.13) is unchanged by these, and since JC is identical for both versions,
so are all the other results in this section.
5.5. The two point function of the graviton 92
As for the gauge bosons, the physical behaviour naturally splits into long distance
k2 I a' « 1 and short distance k2 I a' » 1 regimes. In the former, gravity will be
dominated by the low energy modes, for which the usual corrections to Planck's
constant apply. We can expand the amplitude as a power series in k2 and k2, and
neglect the terms O(k2) relative to O(k2). In the short distance regime however
such an expansion is no longer appropriate, but the amplitude still has terms with
a prefactor of k2 which we should consider dominating over those prefixed by k2. In . .
this way, we may consider the same correlators as being typical dominating terms
in the amplitude for the non-zero B-field corrections to both limits; one such term
is
A :J 100
dt j d2z j d2w - g;~' N't;(8X1-L(z)fJX>.('w)eik·X(z,z)e-ik·X(w,w))
x (k. ~~v(z)k ·1/;1/JP(w)) (5.5.11)
which has a leading contribution of the form
1= dt J d2z J d2w g;N't; 4::a'z(t)(Gt)2kakbBzQaJ.L[Jwgb>.(eik·X(z,z)e-ik·X(w,w))
· L8 + ... . (5.5.12)
Here L8 is the component of this term which contributes to A8, and we have in
cluded in the partition function the Chan-Paton summation, which corresponds to
summing over the Casimirs of the representations of the gauge. group. We shall leave
a complete analysis to future work, and consider the contribution from the corner
of the moduli space where t > 1. Here we can take the derivatives of the Green's
functions to be given by the leading order terms as t ~ oo - as for the gauge theory
case, this is equivalent to a field theory calculation, but it is more expedient to per
form the calculation from string theory. We find that the behaviour is dominated
5.5. The two point function of the graviton 93
by the correlator of the exponentials: this is given by
(5.5.13)
Note that this correctly fadorises onto the corresponding boundary amplitude. 'To
take the field theory limit now, we write T = 1ro/t, y = C:S(z)/t, use the translation
invariance of the annulus to fix C:S(w) = 0, and write ~(z) = x, ~(w) = x', and
insert the partition function and kinematic factors, with a sum over spin structures.
Making use of the identity ( 5.4. 7) and assuming N ~ 1 supersymmetry, so that after
multiplying by the partition function all spin-structure-independent terms vanish,
we obtain the prefactor
(5.5.14)
We shall assume F(T) has the behaviour
lim T 2 F(T) = {3, T-+oo
(5.5.15)
where f3 is a constant. If we now insert the factors from our "typical" contribution,
we obtain
(5.5.16)
As discussed, in contrast to a noncommutative field theory, there is no separation
into planar and non-planar diagrams.
Since we are considering the regime k2 k2 « 1, we reorder the integration and
use the leading behaviour of the Bessel function K 0 as in ref. [78] to give
LFT ~ g2 N2 k2 f3 log lk2k21B2 ( 1 1 + o:'k2 + k2 ) 8 ~ s G 47r2o:' 67r 2' 2 -2- 167r2o:'
~ g2 N2 _}!:_ ./!_ 16~3 a:' log I k2 pI s G 47r2o:' 67r k2
(5.5.17)
5.6. Phenomenology: modification of gravity at a mm 94
and thus this contribution to the graviton renormalisation, after we include Nc =
(8nG4 ) 112j2n (where G4 is Newton's constant) is given by
(5.5.18)
Note that when we sum over all equivalent diagrams and thus remove the field theory
singularity, the log k2 term still remains. However, this is of course not a singularity,
as we have up to this point been considering k » a'.
Ask decreases, the amplitude should smoothly revert to the correction for()= 0.
To find the deviation from Newtonian behaviour at large distances we are interested
in the variation of 8 for small k2 / ci, which as discussed above will be dominated
by the same terms as in the large limit; for the term we have been considering we
obtain
from which we extract the contribution to the renormalisation:
(5.5.20)
5.6 Phenomenology: modification of gravity at a
mm
We now turn to phenomenological issues beginning briefly with the possibility of
Lorentz violation in the photon. In the Introduction we mentioned birefringence of
the trace U(1) photon which is constrained by astrophysical observations. Taking
into account our analysis and the fact that the Lorentz violating operator ll2 vanishes
in a fully supersymmetric theory, the velocity shift is of order
(5.6.1)
5.6. Phenomenology: modification of gravity at a mm 95
Following ref. [94, 95] a relatively firm constraint comes from "time of flight" signals
from pulsars;
..f5.Ms~sYMs rv ~ M~usy < 2 x 10-s, (5.6.2) MNC IR
where A is here a measure of the one-loop suppression in the gauge diagrams, and
Msusy is a measure of the supersymmetry breaking. A natural question to ask
is how low the IR cut-off can be; in other words, is it likely that a regime that is
well approximated by noncom mutative gauge theory will ever be accessible? Alas, the
answer is no. Since A is a loop suppression factor involving known gauge couplings it
will be at least of order 10-3 assuming that the mixing between the physical photon
and trace U(1) photon is of order unity. However supersymmetry is broken and
transmitted, one should almost certainly take Msusy > 1TeV giving
(5.6.3)
This bound is comparable to those coming from atomic physics calculated in ref. [96];
(5.6.4)
Assuming that Ms < Mp1, that bound translates into
(5.6.5)
If the physical photon has significant mixing with the trace U(1) photon, it seems
likely therefore that a non-zero B field would be felt as residual Lorentz violation
rather than full blown noncommutative field theory. For more detailed discussion
of these questions see ref. [72].
Consider instead the possibility that the physical photon does not mix with the
trace U ( 1) photon. This could be the case if the trace U ( 1) photon forms part
of a hidden sector, or if the trace U(1) is spontaneously broken by for example a
Fa yet-Iliopoulos term, if it is anomalous. In this case M NC can be much lower and
a significant effect can show up in gravitational interactions. Our general analysis
shows that the graviton two point function in a theory with non-vanishing () tends
continuously to the commutative one with leading terms suppressed by factors of
k2 /a'. Neglecting the possible implications of a non-trivial tensor structure for the
5.6. Phenomenology: modification of gravity at a mm 96
moment, the mildest effect one expects is a modification of the Newtonian force law
which derives from it. The observable effects will make themselves felt as we probe
the gravitational interaction at shorter distances. As we saw, there is something
akin to a "nonplanar" one-loop contribution in the sense that G(k) interpolates
between the k2 » o:' regime and the k2 « o:' regime where it deviates from the
purely commutative model as k2 / o:'. Neglecting tensor structure, we can therefore
model the two point function as
G(k) = 1 1 + !(~) M 2 k2 1 +A Pl
(5.6.6)
where f ( x) ---t A ( 1 + 0 ( x)) for x « 1 and tends to the short range behaviour for
x » 1. Here M~1 is the one loop Planck mass, which includes also tree level disk
diagram contributions such as those considered in ref. [90]. For example if we assume
that the one-loop contribution has power law behaviour"" lkl(3-p) we can model the
total tree and one-loop two point function as
(5.6.7)
The coefficient A encapsulates the one-loop open string contribution to Planck's
constant in the commutative theory with () = 0, which can be significant and is
model dependent. Indeed there are generic scenarios that lead to the extremes
A » 1 and A « 1:
1. The ADD scenario [97,98]: the Standard Model is associated with a local brane
configuration (for example in a "bottom-up" construction as per the previous
section), with the 4D Einstein-Hilbert action deriving from the dimensionally
reduced lOD action. In this case the one loop correction will be localised
whereas the large tree-level M~1 is the result of a large volume. The one loop
open string contribution will therefore be suppressed by a factor
1 A""-
VlO-p (5.6.8)
where V10_P is the extra-dimensional volume in units of#. 3-branes in the
original ADD scenario with TeV scale gravity would therefore lead to a tiny
5.6. Phenomenology: modification of gravity at a mm 97
>., but one could imagine the Standard Model localised on wrapped D7-branes
for example, in which case intermediate values of ). are possible.
2. The DGP scenario [99]: gravity is localised to a 3-brane in infinite or large
extra dimensions by one-loop diagrams with matter (brane) states in the loop.
The novel feature is that gravity becomes higher dimensional at long distances,
offering an explanation of the observed cosmological acceleration. In this case
one expects >. » l.in the region where gravity is 4 dimensional. In more .detail,
the full action consists of a bulk term and a one loop induced brane term;
(5.6.9)
where R(4) is the curvature form the induced metric on the brane. Since pp-4
appears in the propagator with a factor k2 it is natural that the cross-over
length scale above which gravity appears D dimensional, generically given
by [100] 6-D D-4
R ,_ 2
c =a 4 Pc (5.6.10)
This possibility has been analysed for (Type I) open string models in ref. [92,
93, 100], where in practice a number of different threshold effects are possible if
the matter branes wrap some compact internal dimensions. The precise details
of these other thresholds will not change our conclusions about the effect of
UV /IR mixing.
To see the effect of the one-loop corrections on the potential between two point
particles consider for example 012 = 0. In this case
(5.6.11)
where {) is the angle to the 3 direction. The potential depends on the angle {) and
is given by the retarded Green's function;
(5.6.12)
5.6. Phenomenology: modification of gravity at a mm 98
which leads to
where (} Ms
rc = Jc2 = M'fvc. (5.6.14)
In the limit where r cos '19 » r c we may expand f inside the integral. Using the
identity
(5.6.15)
we find that the leading deviation from Newtonian behaviour is a quadrupole mo
ment that sets in at r rv rc: indeed if f(x) = .\(1 + (3x + ... ) we find
V(r,'!9)= 1 ( 1 +>.(3(3cos2
'19-1)r~+O(r~))· 81r M~1r (1 + >.) r 2 r4
(5.6.16)
The radius rc is the distance above which Planck's constant tends to the B = 0 one
loop value. This is a potential which can be compared directly with the experimental
bounds presented in ref. [101]. Also note that there is a direction given by cos '19 = 0
where the physics is identical to (} = 0 physics.
At smaller distances the "nonplanar" contribution to Planck's constant dimin
ishes. For r « r c we may use the identity
(5.6.17)
and approximate
(5.6.18)
to find the first few harmonics as
V(r, '19) = (5.6.19)
1 ( 1 p-4(-1)nB(p-~-4),n~l) (r)l+n (r)p-2) M2 --, + 2::.:
2 1 Pn(cos'!9) - + 0 - . 81r PIT 1+A n=O n. Tc Tc
The leading term is the tree-level Planck's constant, and the sub-leading terms grow
with radius, as they should, to build up the full one-loop Planck's constant at large
distance.
5.7. Conclusions 99
The most notable general conclusion from this analysis is simply that the distance
scale at which the modification of gravity takes place,
() Ms rc = # = M'Jvc' (5.6.20)
can be much larger than the inherant distance scales in the model. For example if
M8 rv Mp1 and MNc rv 1TeV then3 rc rv 1mm (the same numerical coincidence as
the large extra dimension scenarios with 2 extra dimensions).
5. 7 Conclusions
N oncommutative field theory provides a theoretical framework to discuss effects
of nonlocality and Lorentz symmetry violation. Proper understanding and better
control of the UV /IR mixing has been a serious obstacle for the field theory. In
this work, we have emphasised that the IR singularities are just a reflection of the
fact that field theory is UV divergent. Consequently any attempt to resolve them
without modifying the UV behaviour of the field theory is doomed, and they can
only be consistently smoothed out in a UV finite theory. We have demonstrated
this explicitly by considering noncommutative field theory as an approximation to
open string theory with a background B-field. We showed that the noncommuta
tive field theory description is valid only for the intermediate range of energy scale
AJn = o/ M'fvc < k2 < 1/a' and explored what happens outside this range. The
IR singularities are rendered harmless and in fact, long before they are reached, the
singular IR physics of the noncommutative theory is replaced by regular physics that
is dictated by the UV finiteness of strings. In many non-supersymmetric theories,
tachyonic instabilities arise from the modified dispersion relation (5.1.4) [102], which
our analysis implies are also resolved by embedding into an UV-complete theory, as
discussed in the context of field theory in [72].
With the UV /IR mixing under control, one can now reliably study how noncom
mutative geometry modifies the IR physics. Below the noncommutative IR scale
3 Note that the bound of equation (5.6.4) do not apply since we consider a VEV for a trace U(l)
on a hidden brane, rather than the antisymmetric tensor or the visible electromagnetic field.
5. 7. Conclusions 100
A1 R, normal Wilsonian behaviour is resumed and the low energy physics can be
described in terms of ordinary local physics with residual Lorentz violating opera
tors. Indeed the theory tends continuously to the commutative B = 0 field theory,
with the Lorentz violating operators remaining as a footprint in the low energy phe
nomenology of the string scale physics. A second important example of how the
low energy physics is modified arises in the gravitational sector. We studied how
the noncommutative geometry may modify gravity by considering the graviton two . .
point function. The departure from the ordinary Newtonian potential can be much
more significant and happen at much lower energy scales than those suggested by
any extra dimensions.
One aspect of the present study that requires further elaboration is the nature of
the effective field theory in the gravity sector and the resulting cosmology. Because
of the difficulty of extracting an effective field theory for the gravitational sector it
is not clear how these features will turn out, or indeed if they lead to any strong
observational constraints.
Chapter 6
Summa:ry
This thesis comprises two rather different bodies of work, that both contribute to
the current efforts of string phenomenology.
Chapter 3 examined in detail the calculation of one-loop amplitudes in inter
secting brane worlds. We first used the techniques introduced in previous work [30]
to calculate an amplitude which gives the running of Yukawa couplings, and found
that it contained a divergence. The origin of the divergence was identified as being
a closed string exchange, which indicated that it should cancel provided that the
existing consistency conditions on such models were met. To show that this is in
deed the case required examining the OPE of the boundary-changing operators and
a simple assumption about the behaviour of the amplitudes without a periodic cycle
between the boundaries of the worldsheet, and the Mobius strip diagrams.
The above calculation motivated a clarification and generalisation of the previous
technology for calculating correlators of boundary-changing operators, extending
it to general numbers of insertions and providing the phase matrix Mab essential
for such calculations. It also motivated an extension of the techniques to include
hitherto uncalculable amplitudes, involving the string theory equivalent of loops
without gauge fields. The phase matrix for these processes was also calculated.
While the absolute evaluation of the amplitudes appears to be intractable, it is
certainly possible to obtain useful information from the expressions given here; aside
from confirming the assumption about the cancellation of divergences, it can also
be used for examining the low energy theory, and indeed examining energies near
101
G ' '
:
-\'"':
Chapter 6. Summary 102
the string scale, as illustrated by the analysis of section 3.4. That section examined
the moderation of power-law running of the Yukawa coupling near the string scale,
and argued that the behaviour can be approximated in the field theory by a string
inspired modification of the propagators in the loop.
Chapter 4 provided the necessary theoretical tools for the calculation of instanton
contributions to the superpotential in toroidal intersecting brane worlds. In partic
ular, the Pfaffian and determinant factor was calculated for the first time, including
some new results relevant for gauge threshold corrections, and also some relevant
correlators of spin fields. We then applied these tools to the issue of corrections
to the Yukawa couplings, and found that they can contribute provided that the E2
instanton does not intersect any brane of the model. In this way, they can solve the
"rank one Yukawa problem" from which these models suffer.
Chapter 5 investigated some of the effects of a non-zero antisymmetric tensor field
in string theory models with D-Branes. The infrared singularity in noncommutative
field theories was explcitly shown to be removed when embedded in string theory,
provided that the theory is free of tadpoles. This was illustrated in bosonic string
theory and the superstring. The effect of the antisymmetric tensor upon gravity
was also examined, by computing a portion of the graviton propagator at one loop.
This showed that there is an infrared-ultraviolet mixing in the running of Newton's
constant, but not a separation into planar and non-planar diagrams as might have
been naively expected. The violation of lorentz symmetry creates a directional
dependence for the gravitational potential, which would be a clear experimental
signal and in principle possible to observe at distances as large as a millimeter.
It is hoped that the ideas contained within this thesis, in addition to solving
some interesting problems within the field of string phenomenology, present several
avenues for further insights into string theory.
Appendix A
Material Regarding Intersecting
Brane Worlds
A.l Boundary-Changing Operators
The technically most significant part of any calculation involving chiral fields stretched
between branes is the manipulation of the boundary-changing operators. They are
primary fields on the worldsheet CFT which represent the bosonic vacuum state in
the construction of vertex operators, so we require one for every stretched field. In
factorisable setups such as we consider in this paper we require one for every com
pact dimension where there is a non-trivial intersection- so we will require three for
each vertex operator in N = 1 supersymmetric models.
In previous papers (e.g. [15, 16]) the boundary-changing operator for an angle ()
has been denoted a8 , in analogy with closed-string twist-field calculations. It has
conformal weight h8 = ~B(1- B), and we can thus write the OPE of two such fields
as
av(zi)a>.(Z2) rv L c~>,ak(z2)(z2- zl)hk-hv-h>. (A.1.1) k
Moreover, the "twists" are additive, so k = v + A for v + A < 1, or k = v + A - 1
for v + A > 1. However, they also carry on the worldsheet labels according to the
boundaries that they connect; for a change from brane a to b we should write a~ab).
We then have non-zero OPEs only when the operators share a boundary, so that we
103
A.l. Boundary-Changing Operators 104
modify the above to
(A.1.2)
We also find that new fields appear, representing strings stretched between paral
lel branes; the OPE coefficients and weights can all be found by analysing tree-level
correlators. In order to normalise the fields and thus the OPEs we must compare
the string diagrams to the low-energy field theory, specifically the Dirac-Born-Infeld
action for the I I A theory:
S6 = -T6 J d7~e-il>trJ- det(G + B + 2na'F) (A.1.3)
where G and B are pull-backs of the metric and antisymmetric tensor to the brane
world-volume, F is the gauge field strength, and eil> is the closed string coupling.
If we consider gauge excitations only on the non-compact dimensions, the effective
field theory has gauge kinetic function [15, 32]:
-2 -il> 1 rr3 L~
9D6 = e 27r 2 G "'=1 7ry a·
(A.1.4)
where we have the volume of brane a equal to La = TI!=1 L~. If we now consider
two, three and four-point gauge-boson amplitudes (for non-abelian gauge group) in
the 4d effective theory, they must all have the same coupling; this is only possible
if the gauge kinetic function is associated to the normalisation of the disc diagram,
and not the fields, implying
( ')-1 -2(1) = -2 a 9o a 9a (A.1.5)
where g0 is the open string coupling. Clearly the gauge bosons must be canonically
normalised by a factor of g;; 1 to match the field theory, where the coupling is also
absorbed into the fields. Note that the above can also be obtained by a boundary
state analysis without recourse to the low energy supergravity [103, 104].
We now consider the two-point function for boundary-changing operators; we
require
( ab ba ) G c<aba)O (1) (]"8 (]"1-8 = CoCo = 8,1-8 a (A.1.6)
A.l. Boundary-Changing Operators 105
where Gc9c
0 is the Kahler metric of the chiral multiplets [32]. This allows us to
determine the OPE coefficients; however, for convenience we shall combine them
(for now) with the string coupling since that is how they shall always appear:
3
( ) 2 IT c(aba)O ( ')-1 2G 9o 0"',1-0" = a 9a CabGba (A.l.7)
K:=1
To obtain further correlators we must now consider four-point tree diagrams. First
we consider Atree- g~(a~b(z1 )ar~1)z2)af~),(z3 )a~a(z4 ))n2 , which was calculated for
a particular limit in [15, 32], but for the general case in [16] (note that the product
over dimensions is implied). The result was
3
Atree = C(a')-1g; II z122hvl< z"342h>.~< JK:(x)-112 (1- x)-v"..\" K:=1
where v21 (n1) = L~1 'K; + n1Ll; and v32 (n2) = L~2'K; + n2L~11 depend on the configura
tion; L~1 'K; is the distance between the intersections 1 and 2 etc; x = (z12z34/ z13z24 ) is
the only actually independent coordinate; and the various hypergeometric functions
are given by:
Ff(1- x)
F;(l- x)
K~(x)
K;(x)
rK;(x)
JK:(x)
F(1 -vii; 1 - ..\K; 2- vii; - ..\K; 1 - x) ' ' '
F(vK, ..\\ 1, x)
F(1- vii;, 1- ..\K;, 1, x) = ·(1- xt"+..\"-1 K1(x)
(1
_ x) 1-v"-..\" B(1- v, 1- ..\K) F2(1- x) B (vii;, 1 - vii;) K 1 ( x)
B(1- v, 1- ..\K;) F2(1- x) B(vK;, 1- vii;) K 2(x) B (vii; , ..\ K;) F1 ( 1 - x)
B (vii;, 1 - vK) K 1 ( x) (1- x)1-v"-..\" B(vK;, 1- vK;)K~(x)K;(x)(rK; + rK') (A.l.9)
The manifestly-S£(2, C)-invariant form of the above is obtained by premultiplying
the amplitude by (z4 ) 2h>- and taking {z1 , z2 , z3 , z4 } ~ {0, x, 1, oo }. To link the above
expression with that in [15], we must put d'2 = v3'2, d~ = d'2 + (3v21 , where
(A.l.lO)
A.l. Boundary-Changing Operators 106
In the limit considered there, they set d2 = d]T jT'. However, we continue with the
above expression, and first normalise it by considering the limit x -----+ 0, i.e. z2 -----+ z1 ,
Z4 -----+ z3 . This gives
3
Atree(X-----+ 0) rv (a')- 1 g;(O'~aaii)(O)O'~alla)(1))a II c~~~~~~ .. c~~~{~:~" (A.l.ll) K=l
where we now have stretch fields for strings stretched between parallel branes. In
the effective field theory, these are highly massive for large separation, and thus
non-supersymmetric; we normalise them for consistency with the case all = a to
give (1)a, in which case we obtain
(A.l.12)
Applying this to our expression, in this limit TK' -----+ T -----+ - si: 1rv" ln x, and so we the
instanton sum over n 2 vanishes- requiring us to Poisson resum as in [15], giving
3 ~/ 2
A ( ) ( ') 1 2 II -2h .. -2h;.. .. v L;7ra· ~ tree X -----+ 0 rv c a - 9o zl2 v z34 LK X4rr n'
K=l . a
(A.l.13)
where yK is the perpendicular distance between branes a and all in sub-torus /'i,; the
zero-mode of v32 is irrelevant. This gives us two pieces of information: first, we
obtain the normalisation
<I> e ( )5/2 C- -G - G - 21r - 'g2 CabCba CacCca a o
(A.l.14)
(almost) in agreement with the similar case considered by [15]. Secondly, we obtain
II (Y")2 the conformal weight of the operators O'aa : hK 11 = -
2 2 , • aa 1r o
Now we must determine the more general coefficients by taking the limit z3 -----+ z2 ,
or equivalently x -----+ 1. In this case TK -----+ 0, TK' -----+ {3, and we obtain
(A.l.15)
where
l/K + AK < 1 YK= (A.l.16)
l/K + AK > 1
A.2. One-Loop Amplitudes With Gauge Bosons 107
and A~~:(n1, n2) is the sum of the areas of the two possible Yukawa triangles formed
by the intersection of the three branes wrapping in both directions, given by
v( v) _1 sin 7rvll: sin 7r _\II: (( v )2 ( II: (31\; II: )2) A~ n~~:1 ,n2~ = v" + v +v 2 sin n(1 - v~~:- >,~~:) 32 21 32
~ sinnv~~:sinn>.~~: ((d~~:) 2 (n~~:) + (d~~:) 2 (n~~:)) (A.1. 17) 2 sin n ( 1 - v~~: - >. ~~:) 1 1 2 2
(and a similar expression for >,~~: + v~~: > 1). With this expression, it becomes imme
diately clear how we can obtain the OPE coefficients:
~~:=1
and thus we infer
3
#g e-if>/2 II c(cab)v~<+).l< = (2nGCbcCcbG - G - ) 1/2 o v~< ,.>.~< CabCba CacCca
~~:=1
3
II ( vz;;:y~~:)1/2 2::= exp K=1
A~~:(m~~:)
2na'
(A.1.18)
(A.1.19)
where now A~~:(m~~:) is the area of the triangle bac, while the conjugate coefficient
will have the area of ballc (zero if all = a). This concludes the determination of all
the relevant OPE coefficients in the theory. For use in the text, we define
3
y(cab) = g II cCcab)(G - )1/2 0 CbcCcb (A.1.20)
~~:=1
and yballc analogously; they represent the physical Yukawa couplings.
A.2 One-Loop Amplitudes With Gauge Bosons
A.2.1 Classical Part
The prescription of [57] applies to the evaluation of the classical part of boundary
changing operators, by the analogy with twist operators, after we have applied the
doubling trick- the nett result being that we must halve the action that we obtain.
Since it is the only consistent arrangement for two and three point diagrams, and
the most interesting for four-point and above, we shall specialise to all operators
A.2. One-Loop Amplitudes With Gauge Bosons 108
lying on the imaginary axis, where the domain of the torus (doubled annulus) is
[-1/2, 1/2] x [0, it]; this provides simplifications in the calculation while providing
the most interesting result. In the case of all operators on the other end of the
string, we would simply define the doubling differently to obtain the same result,
and since the amplitude only depends on differences between positions the quantum
formulae given would be correct.
For L operators inserted, we must have M = l::f=1 Bi integer to have a non
zero amplitude; for vertices chosen to lie on the interior of a polygon we will have
M = L- 2, which h shall be the case for the three and four point amplitudes we
consider - in the case of a two-point amplitude we must have M = 1. Labelling the
vertices in order, we denote the first L - M vertices { Zo:}, and the remaining M
vertices {z13 }, we then have a basis of L- M functions for BX(z) given by
L-M L
wo: = B1(z- zo:- Y) IT B1(z- zj) IT B1(z- zk)lh-1 (A.2.1) jE{o:}#o: k=1
where
(A.2.2) i=1 j=L-M+l
and the basis of M functions for BX(z)
L L
w13 = B1(z- ZJ3 + Y) IT B1(z- Zj) IT B1(z- zk)-0k (A.2.3)
jE{/3}#/3 k=l
We are then required to choose a basis of closed loops on the surface. There are
two cycles associated with the surface which we shall label A and B, and define as
follows:
1 dz ~A 1 dz ~B
1-1/2
dz tt-1/2
11/2
dz -1/2
(A.2.4)
The remaining integrals involve the boundary operators, and we define (not-closed)
loops ci by
1 1Zi+l
dz = dz C; z;
(A.2.5)
A.2. One-Loop Amplitudes With Gauge Bosons 109
Note that we have defined in total L-1 C-loops, and they are actually linearly depen
dent, since we can deform a contour around all of the operators to the boundary to
give zero- we only require L-2. These could then be formed into (closed) Pochham
mer loops by multiplication by a phase factor, but it actually turns out that this is
not necessary. We form the above into a set {!'a}= {J'A, /'B} U {Ci, i = 1, ... , L- 2},
and define the L X L matrix w~ by
w: 1 dzwa(z)
""''a
w! - ~ dzw13 (z) }ia
(A.2.6)
The boundary operators induce branch cuts on the worldsheet, which we have a
certain amount of freedom in arranging. We shall choose the prescription that the
cuts run in a daisy-chain between the operators, with phases exp(iai) when passing
through the cut anticlockwise with respect to Zi defined as follows:
a1 2nB1
QL-1 ~2nBL
2n(}i Qi- Qi-1
Qi 2n L (}j (A.2.7) j=1
noting that ai is only defined modulo 2n. Each path I' a is associated with a physical
displacement Va (which shall be determined later); i.e
(A.2.8)
and, since we can write ax and ax as linear combinations of wa and wf3 respectively,
we obtain
(A.2.9)
where the sum over primed indices is understood to be over {a}, and over double
primed indices to be over {,6}, and for reference we have used the complexification
X = ~(X1 + iX2). The inner product is defined to be
(wi', wj') = j d2zwi' (z)wj' (z) _ iW~'w[ Mab (A.2.10)
A.2. One-Loop Amplitudes With Gauge Bosons 110
and similarly
( i" j") _ 1 Jd2 i"( )-j"(-) _ ·- i" j"M-ab w 'w - -- zw z w z = 'l wa wb 41fi:Y'
(A.2.11)
and thus .tVfab =-Mba. As in [57] we have MAE= -MBA= 1, but, after performing
the canonical dissection on the torus for our arrangement of branch cuts and basis
loops we determine:
Mmm
2i ei "'t-;"'m sin (aL--: 1-a1) sin Qm m < l · ("'L-1) 2 2 Slll -2
-
2i Sin (ll<£-1-ll<m) Sin ll<m · ("'L-1) 2 2 Sill -2
-
(A.2.12)
Inserted into the expression for the action, this gives S = 4;a' Va vbsab, where
(A.2.13)
At this point we determine the displacements. Clearly, since the boundary
Re(z) = -1/2 contains no boundary-changing operators, along 'YA the string end
is fixed to one brane, which we shall label a. Thus, VA represents the one-cycles
of brane a, and is equal to ~nALa, where La is the length of a and the factor of
1/V'i is due to the complexification we chose. Now, the technique that we are using
requires there to be a path around the boundary of the worldsheet where we do not
cross any branch cuts: if the string has one end fixed on brane a along path 'Y A, the
other end (at z = 0) must either reside on brane a or a brane parallel to it, which
we shall denote all. Path 'YB is related to the displacement between these branes.
Consider the doubling used, for coordinates aligned along brane a:
{
ax(z) ~(z) > 0 ax(z) = --
-aX( -z) ~(z) < o (A.2.14)
and a similar relationship for -ax( -w) and ax. Note that we have .6./Ax = .6./ax,
in keeping with our identification of VA· We have
.6.18 = 1 dzaX + 1 dz8X(z) 'YB /B
11/2 d d -dx-X(x)- -X(x)
0 dx dx
ivi2(X2(a)- X2 (a11))
. f2( nB47r2T2 )
ZVL. L +y a
(A.2.15)
A.2. One-Loop Amplitudes With Gauge Bosons 111
where y is the smallest distance between a and all and T2 is the Kahler modulus of
the torus, defined earlier. This resolves an ambiguity in [30]. The remaining paths
have straightforward identifications, since they are essentially like the tree-level case;
they are the displacements between physical vertices. We must identify each portion
of the Re(z) = 0 boundary with a brane segment between intersections; so for the
case
(A.2.16)
we have v1 = ~(n1Lb+L12 ) and v2 = ~(n2Lan +£23). If we were to use Pochham
mer !oops for these, we would multiply these by Pochhammer factors - but these
would be cancelled in the action by those associated with the loops on the world
sheet.
Note that although we are free to choose the arrangement of branch cuts on the
worldsheet, we have no freedom in identifying the displacement vectors, and thus
the daisy-chain prescription is the simplest, particularly since we are restricted in
the permutations of the boundary-changing operators - we must ensure that each
change of brane has the correct operator to mediate it.
A.2.2 Quantum Part
The quantum part of the correlator is given in terms of the variables defined in the
previous section as
L L-M
(IJ O"en(zn)) = IWI-112BI(Y)L-2 IT B1(zi- Zj)
n=l i<j L-M<i<j
L IJ (h(zi- zi)-B;Bj-(l-B;)(l-Bj) (A.2.17)
i<j
where IWKI is the determinant of the matrix of integrals of cut differentials. Note
that this expression does not depend upon the brane labels of the boundary-changing
operators, as it is not sensitive to the particular boundary conditions.
The correlators for the fermionic component of the vertex operators do not receive
worldsheet instanton corrections, so they are given by the relatively simple formula
A.2. One-Loop Amplitudes With Gauge Bosons 112
presented in [30]:
L L (II eiq;H(z;))v = Bv(L QiZi) II (;11 (zi - Zj )q;qj (A.2.18) i=1 i=1 i<j
the above is a generalisation of the formulae for standard spin-operator correlators
(see e.g. [37]) and is thus also valid for the 4d spin correlators.
A.2.3 4d Bosonic Correlators
Beginning with the Green's function for the annulus, obtained via the method of
images on the torus:
' . ' ( ) a I ( ) 12 7rZa O< 2 G z, w = - 2 ln 01 z- w - -t-(~(z- w))
' . ' a I 2 7rZa O< 2 - -ln th(z + w)l - -(~(z + w)) 2 t
(A.2.19)
and specialising to all points at boundaries (i.e. z = -z or 1 - z), we obtain the
amplitude on the annulus
M
Ai - (II eik;·X(z;))
i=1
(A.2.20)
where
(A.2.21)
is the partition function for the non-compact bosons with the be-ghost contribution
included. We then use the green's function to obtain
4
(8XJ.!3(z)8XJ.!4(w) II eik;·X(z;)) =(II eikm·X(zm))
i m=1
4 4
{ 2ry"" a.awG( z' w) + 4 ~ ~ kf' kj' a.G( z, z,)awG( w' Zj)} ( A.2.22)
which is required for 4-point functions.
A.3. One-Loop Amplitudes Without Gauge Bosons 113
A.3 One-Loop Amplitudes Without Gauge Bosons
A.3.1 Classical Part
Cut Differentials
The cut differentials for diagrams on the doubled annulus where there exists no
periodic cycle necessarily have modified boundary conditions. Where the diagram
with no boundary changing operators is the partition function of strings stretched
between branes a and b with angle 7r0ab, the conditions for the one form ax(w) are
(see section 2.4.2):
while
8X(w +it)= 8X(w)
8X(w + 1) = e2niBab8X(w)
8X(w +it)= 8X(w)
a.X(w + 1) = e-2niBabaX(w)
(A.3.1)
(A.3.2)
whereas the local monodromies remain the same as for the periodic case. Indeed,
we can retain many of the elements of those differentials, including L
!x(z) = II 81 (z- zi)B;-l
i=l
L
lx(z) = II 81 (z- zi)-B; (A.3.3) i=l
These satisfy the local monodromies; to construct a complete set of differentials
satisfying the global monodromy conditions, we note the identities:
[ 1/2 +a ] [ 1/2 +a ] () (z + m; r) exp(27ri(1/2 + a)m)O (z; r)
1/2 . 1/2 (A.3.4)
() (z + mr; r) exp( -1rim- 1rim2r- 21rimz)O (z; r) [
1/2+a] [ 1/2+a]
1/2 1/2
which show that we only need to modify one theta function from the periodic case;
we have also
0 [ c: a ] (z; T) ~ exp[Z1ria(z +c) + a21riT]B [ : ] (z +aT; T) (A.3.5)
A.3. One-Loop Amplitudes Without Gauge Bosons 114
which is crucial for showing the equivalence between the approach that we are about
to use, and the method of obtaining these amplitudes by factorising higher-order
amplitudes calculated by the previous method. Denoting the theta-function
we construct the set of L - M differentials for 8 X ( z), similar to before
L-M
w~ab(z) = l'x(z)B+ab(z- Za- Y) IT el(z- Zj) jE{a}""a
where Y is as defined before; and we have the set of M differentials for oX L
w~ab(z) = l'x(z)B_ab(z- Zf3 + Y) IT el(z- Zj) jE{/3}""/3
(A.3.6)
(A.3.7)
(A.3.8)
We demonstrate that these are complete sets in the same way as in [57]: first,
note that the function are independent, since wa(z#a) ..:..._ 0. Then suppose we had
another differential w'(z); we construct the doubly-periodic meromorphic function
>.(z)
>.(z) = w'(z) - ~ ci wi(z) wl(z) i=l wl(z)
(A.3.9)
At z = z1 or z E {z13 }, the above is not singular, while we can adjust the L- M
constants Ci to cancel the residues of the poles at z = zi""l and z 1 + Y- Babit. Thus,
since ). has no poles, and it is doubly periodic on the torus, it is a constant(the last
point contains the only subtlety with respect to the earlier case; even though the
differentials wa are not periodic on the torus, because they only acquire a phase, the
differential ). is periodic). Hence, since C1 just multiplies a constant, we can adjust
it to set ). to zero. The same follows for wf3.
Note that if we want to use the same theta-functions throughout, we could choose
a basis obtained by factorising the functions used earlier. In this case, we obtain
L-M
w~~b(z) = e1ri(l-it)Oabe27TiBabZI'x(z)Ol(z- Za- y + Babit) IT el(z- Zj) (A.3.10) jE{a}""a
and L
w~ab(z) = e-1ri(l-it)Babe-21riOabZI'x(z)Bl(z-z{3+Y -Babit) IT el(Z-Zj) (A.3.11)
jE{/3}""/3
A.3. One-Loop Amplitudes Without Gauge Bosons
The relation between the bases is given simply by
w~~b(z)
w~'ab(z)
ent(Oab+O~b) e2ni0ab(za+Y) Wo: (z) +ab
e -nt(Oab+O~b) e -2ni0ab(Z,J-Y) W{3 (z) -ab
115
(A.3.12)
The choice of basis is not important for the following section, and the amplitude
will of course be independent of the basis choice.
Canonical Dissection
With our basis of cut differentials for the doubled annulus we require their Hermitian
inner product to calculate the action. This is given by
(wi,wj) = r d2 zwjwi = i 1 wj(z)dz r dzwi J R hm lzo
(A.3.13)
This is the canonical dissection of the surface, where the contour passes anticlockwise
around the surface without crossing any branch cuts. The task is then to choose the
most convenient arrangement of cuts, and express the above in terms of integrals
of paths on the surface corresponding to physical displacements. Since we consider
operators only on one boundary, we arrange them to be on the imaginary axis; since
we always have one vertex fixed, we choose this to be ZL and place it at the origin.
The integrals between vertices are then defined by
where <s(zn) > 8'(zn+1 ). We also have the loops
and the phases
1-l 1-1/2 -l+it - -l/2+it
L:
27r 2: ()j - 21r()ab j=l
(A.3.14)
(A.3.15)
(A.3.16)
A.3. One-Loop A1pplitudes Without Gauge Bosons 116
we also label the additional cycle /D:
lit
ZI
-21T()ab (A.3.17)
which we eliminate, along with cl' via the equations
n=2
L-l
eiaD/D + eiaiCl + L eianCn = -!A (A.3.18) n=2
Hence we have chosen our set of paths to be {!A,/B} U {C2 .. C£-1}. With these
definitions, we now perform the canonical dissection, and defining our matrices as
before W~ =fA wi' etc, the inner product is of the form
and the results are
1 Sinal Sin aD
2i 2 2 Sin aD-al
2
. e-ian . (an-al) . aD -22 . (a -a ) sm sm -sm~ 2 2
2i i~. (am-aD). (a.n-al) -----,-e 2 sm sm ---sin( aD~al) 2 2
(A.3.19)
(A.3.20)
where m ~ n in the last line, L - 1 ~ n, m ~ 2 and the elements reflected in
the diagonal can be obtained from the above using Mdc = - !Vfcd. We see that we
can obtain appropriate expressions for the previous case simply by taking Bab = 0,
in which case the formulae are greatly simplified, with the AA and An-elements
vanishing. The above expression then yields the classical action by equation (A.2.9).
A.3.2 Quantum Part
As for the classical part, the quantum part of the bosonic correlator for these dia
grams may be determined in two ways; factorisation of a diagram with a larger num
ber of operators inserted, or by calculating new green's functions and proceeding by
A.3. One-Loop Amplitudes Without Gauge Bosons 117
the stress-tensor method. However, unlike for the classical part, it is straightforward
to perform the factorisation.
To obtain the quantum amplitude for L boundary-changing operators with angles
{ 6q and overall periodicity Bab, we start with an amplitude with L + 1 operators
with angles {Bab,()i,eL- Bab} which we assign to vertices at {z0,zi,O}. The above
choice of angles ensures that in both sets of equations M is the same, and we have
(A.3.21)
The full quantum correlator is
L-M L
Zqu = f(it)IWL+11- 112B1(YL+1)(L-1)/2 II e1(Zi- Zj)112 II B1 (zi- zj) 1/
2
O<Si<j L-M<i<j
L
II e1(zi- Zj)-![!-O;-Oj+ 20;0j] (A.3.22) O<Si<j
where f(it) is the normalisation. We then note that in the limit z0 ----+it, all.except
two of the integrals in the matrix W L+ 1 are finite. Indeed, only w0 ( z) develops a
singularity, and only the integrals of it over the cycles 'YB and C0 become infinite
(note that we are using the set of curves {"(A, !B, C0 .. CL-d ). The determinant
becomes in the limit
(A.3.23)
where IWfl is the determinant with the 'YB cycles and w0 integrals deleted: we then
note that the rows are not linearly independent due to the identities (A.3.18), and
it is therefore zero. We must finally evaluate (WL+I)g.
L-M L-1 ()1 (-YL+1) II ()1 (it - zi)0
i II e1(it-zj)0i-
1 (A.3.24) i=1 j=L-!vl+1
which, when we consider that the amplitude should factorise according to
(A.3.25)
A.3. One-Loop Amplitudes Without Gauge Bosons 118
we find that the quantum portion of the amplitude is
L-M
z~~b) = g(it)IW{I-l/2e27fiPel(YL- Oabit)(L-2)/2 II Bl(Zi- Zj) 112
O<i<j
L L II 01(zi- zj) 112 IT 01(zi- zj)-~[l-O;-Oi+20;0j] (A.3.26) L-M<i<j O<i<j
where now WL contains integrals of the primed basis of cut differentials, and
(
L-M L )
P ~ Y + ~ ~ z; - ;~J;1+ 1
z; (A.3.27)
Now we find that the "natural" basis for these functions is that which we defined in
equations (A.3.7) and (A.3.8); in this basis, using the relations (A.3.12) we find the
amplitude to be
L-M
z~~b) = g(it)IWLI-1120-ab(YL)(L-2)12 II el(Zi- Zj) 112
O<i<j
L L
II e1(zi- zj) 1; 2 II 01(zi- zj)-W-o;-Oj+29i9il (A.3.28)
L-M<i<j O<i<j
The above can also be obtained by repeating the analysis of [57]; most of the
steps are the same, since the quantum amplitude does not depend upon the exact
form of the greens' functions, only certain constraints upon them, which remain the
same for these amplitudes.
A.3.3 Fermionic Correlators
The fermionic correlators for these amplitudes are easily obtained by simply fac
torising equation (A.2.18); we obtain
L L (II ei(O;-l/2)H(z;))v,Oab = e21fi(Oab-l/2)POv(it(Oab- 1/2) + L qizi)
i=l i=l
II Bl(Zi- Zj)(O;-l/2)(9j-l/2) (A.3.29) i<j
A.4. Theta Identities
A.4 Theta Identities
Throughout we use the standard notation for the Jacobi Theta functions:
B [ : ] (z; T) ~ n~= exp [ 1ri(n + a)2
T + 21ri(n + a)(z +b)]
and define Oaf3 - () [ a/2
] , so that we have the periodicity relations {3/2-
exp(2?riam)B [:] (z;T)
119
(A.4.1)
(A.4.2) B [ : ] (z + m; T)
0 [:] (z+mT;T) exp(-21ribm) exp( -1rim2T - 2?rimz )B [ : ] ( z; T)
We also define, according to the usual conventions,
(A.4.3)
Appendix B
Material Regarding Instanton
Calculations in Intersecting Brane
Worlds
B.l 4d Spin Field Correlators
In the evaluation of annulus contributions to the superpotential involving two fermionic
fields and two supersymmetry modes, it is necessary to evaluate the correlator of
four left-handed spin fields in four dimensions. In general, there may also be picture
changing operators in the amplitude, although there are not for the particular case
in section 4.3. The calculation is performed using the techniques of [105]; the proce
dure is to construct a complete set of Lorentz structures and determine their coeffi
cients by finding particular values for the spinor /Lorentz indices for which only one
structure is non-zero, and evaluating the correlator in those cases. For the general
case when there are two picture-changing operators inserted on the non-compact
directions for four like-chirality spinors { u1, u2, u3 , u4}, the amplitude is given by
4
(w 11 (z)wv( w) IT ( ui)a;§a; (zi)) = -G( u2u4)(u3Cf113 f 114u1)+J( u1u2)(u3Cf113 f 114 u4) i=l
120
B.l. 4d Spin Field Correlators 121
G (S2(z1)S2( z2)S2(z3)S1 (z4)w0 (z) w1 ( w))
J (S1 (zi)S2(zz)S2(z3)S2(z4)w0 (z) w1 ( w))
H (S2(z1)S2(z2)S1 (z3)S2 (z4)w0 (z) w1 ( w))
B (S2(z1)S1 (z2)S1 (z3)S2 (z4)\lf0(z )w0( w))
c (S1 (z1)S2(z2)S1 (z3)S2(z4)\lf0 (z )w0( w)). (B.l.2)
Note that we have written the functions in terms of gamma matrices rather than the
standard Weyl-notation matrices aJ.Lv since the amplitude with PCOs is summed over
momenta - and it is then possible to cancel many terms via the on-shell conditions.
We sacrifice obvious antisymmetry on the inserted operators, but it is straightfor
ward to show that it is still antisymmetric on exchange of '1j;J.L3 ( z) and 'lj;J.L4 ( w). To
demonstrate that the above is a complete set, we require the standard Fierz iden
tities, but we also require corresponding identities among the products of Jacobi
Theta functions. In particular, we require
- B1(z1- z2)B1(z1- z3)81(z2- z3)B1(z- z4)B1(tv- z4)B1(z- w)
O ( Z4 - Z1 - Z3 - Z2 )O ( Z3 + Z2 - Z4 + Z1 _ ) v
2 +z v
2 w
- B1(z1- z4)B1(z2- z3)B1(z- z2)01(z- z3)B1(w- zi)B1(w- z4)
O ( Z1 + Z4 - Z2 - Z3 )B ( Z2 + Z3 - Z1 - Z4 _ ) v
2 v
2 +z w
= 0. (B.1.3)
To prove this, it is easy to check that the periodicities of the terms are the same,
and when one of the functions is zero, the remaining three sum to zero. Thus we
can write any one of the functions as a constant multiple of the other three; since
we can do this for any function the constant must be -1, so the identity holds in
general.
B.2. Spin-Structure Summation 122
The reader may then substitute Baa ,(Ja4 for ( u3)aa, ( u4)a4 • This results in many
simplifications, because we need only keep structures involving B1 B2 = -B2B1. How
ever, to obtain the amplitude without PCO insertions, we can use the OPE of the
\ll fields in the above. Alternatively, we write the amplitude as
where
4
(II (ui)a;§a; (zi)) = A1 (ul u3)( u2u4) + A2(u1u4)( u2u3) i=l
Here we have used
(u1u2)(u3u4)
(S1(z1)S1 (z2)S2(z3)S2(z4)).
(B.1.4)
(B.l.5)
(B.l.6)
Substitution of Baa, Ba4 for ( u3)aa, ( u4)a4 and disregarding terms proportional to Bi
and B~ results in the expression given in equation 4.3.2.
B.2 Spin-Structu~e Summation
It is possible to compute the spin-structure summation for the expression 4.3.2, since
the non-compact spin structure is partially cancelled by the spin~dependent part of
the superconformal ghost amplitude. The result is
(vo vl/2vl/2v;-l/2v;-l/2) -+. ·'· caf3.J, B B J d <l>ab 1/Jbc 1/Jca (} (} a,E2 = lf'(ab) 'f/(bc)a 'f/(ca){3 1 2 t
3 it g 1 dzi f(z2- z3, t) Xj--->ZJ,x~~2.xa--->za (xi- zi)(x2- z2)1/2(x3- z3)1/2Bl(z2- z3)1/4
3 . 3 {2 -(II etk;·X(z;)) 2:::.= II y ~(8X'"(y~e)a</>~b(zi)a</>bJz2)arf>:~Jz3)) t=l {Yl,Y2,Ya}=P(xi,X2,xa) ~e=l
( eiH"'(y"') ei(rf>(ab) -l)H"'(zi) ei(</>(bc) -l/2)H"'(z2) ei(rf>(ca) -l/2)H" (za)) (II indep.)
B.2. Spin-Structure Summation 123
where N
(IJ eia;H(z;))(v indep)- II Bl(zi- Zj)a;aj (B.2.2) i=l i<j
and
encodes all of the dependence on the position of the V0 insertions. It is an odd - -
function of 6, and hence the amplitude does not have a pole at z2 = z3 , as required.
Appendix C
Field Theory -Limits of String
Diagrams - a Review
First we divide the Schwinger integrals as described above so that
( C.l.1)
where UV and IR indicate t E [0, 1] and t E [1, oo] respectively. Considering the I R
contribution of the planar diagram note that, if we reduce () -+ 0, then the field
theory limit should be the same for planar and non-planar diagrams; equivalently,
they should be the same up to 0( a') corrections. This is not immediately obvious
from the Green's functions, but we must bear in mind that the planar diagrams
have spurious poles on the worldsheet, and the string amplitude is strictly only
defined after analytically continuing the momentum [82, 85]. The field theory limit
is obtained by taking t » 1 and excising the regions around the poles - i.e. the
region lx- x'l < 1 and t- lx- x'l < 1 - and then keeping terms of lowest order in
w:
(C.1.2)
where ~ is of order w, and the -( +) preceding it applies to planar (non-planar)
diagrams. We retain this term due to the presence of the tachyon, as in [78, 82]; it
is given by
(C.l.3)
124
Chapter C. Field Theory Limits of String Diagrams - a Review 125
so that A2 = -4n2 +0(w), but for the superstring we shall find that it is irrelevant.
Inserting the above into (5.3.10) and extracting the contribution from the first level
in the loop, we find
This result looks just like the field theoretical Schwinger integral as it should (note
the change to the parameters T = 2na't and y = xjt). We have not explicitly
written the tachyonic contribution or contributions coming from states at higher
excitation level: the tachyon because it is unphysical, and the higher states because
their nonplanar counterparts in the IR (p -+ 0) are all finite. For the moment we
need only note that a contribution at level n yields a Schwinger integral of the form
{oo dTT- <P;l> t dy (1- 2y?e-T(k2(y-y2)+(n-l)o:'-1).
12-rra' Jo (C.1.5)
To obtain the field theory limit, we perform the integrals above and then take
the o:'-+ 0 limit; we can do this using the exponential integral. For example, when
p = 3 we have the standard field theory behaviour, with
(C.1.6)
For d = 22 we obtain the beta function of ref. [78], but for the case d = 0, we find
that the leading logarithm cancels, and we have the finite result
(C.1.7)
Bibliography
[1] J. Polchinski, "String Theory", Cambridge University Press (1999).
[2] P. Di Francesco, P. Mathieu and D. Senechal, "Conformal Field Theory",
Springer ( 1999).
[3] R. Blumenhagen, M. Cvetic, P. Langacker and G. Shiu, "Toward realistic
intersecting D-brane models," arXiv:hep-th/0502005.
[4] R. Blumenhagen, B. Kors, D. Lust and S. Stieberger, "Four-dimensional
string compactifications with D-branes, orientifolds and fluxes," arXiv:hep
th/0610327.
[5] R. Blumenhagen, L. Gorlich and B. Kors, "Supersymmetric 4D orientifolds
of type IIA with D6-branes at angles," JHEP 0001 (2000) 040 [arXiv:hep
th/9912204].
[6] M. Cvetic, G. Shiu and A.M. Uranga, "Chiral four-dimensional N = 1 super
symmetric type IIA orientifolds from intersecting D6-branes," Nucl. Phys. B
615 (2001) 3 [arXiv:hep-th/0107166].
[7] E. Kiritsis, "D-branes in standard model building, gravity and cosmology,"
Fortsch. Phys. 52 (2004) 200 [Phys. Rept. 421 (2005) 105] [arXiv:hep
th/0310001].
[8] A. M. Uranga, "Chiral four-dimensional string compactifications with in
tersecting D-branes," Class. Quant. Grav. 20 (2003) S373 [arXiv:hep
th/0301032].
126
Bibliography 127
[9] M. Cvetic and I. Papadimitriou, "More supersymmetric standard-like models
from intersecting D6-branes on type IIA orientifolds," Phys. Rev. D 67 (2003)
126006 [arXiv:hep-th/0303197].
[10] D. Lust, "Intersecting brane worlds: A path to the standard model?," Class.
Quant. Grav. 21 (2004) S1399 [arXiv:hep-th/0401156].
[11] C. Kokorelis, "Standard models and split supersymmetry from intersecting
brane orbifolds," arXiv:hep-th/0406258.
[12] R. Blumenhagen, "Recent progress in intersecting D-brane models," Fortsch.
Phys. 53 (2005) 426 [arXiv:hep-th/0412025].
[13] A. M. Uranga, "Intersecting brane worlds," Class. Quant. Grav. 22 (2005)
S41.
[14] R. Blumenhagen, M. Cvetic, F. Marchesano and G. Shin, "Chiral D-brane
models with frozen open string moduli," JHEP 0503 (2005) 050 [arXiv:hep
th/0502095].
[15] M. Cvetic and I. Papadimitriou, "Conformal field theory couplings for inter
secting D-branes on orientifolds," Phys. Rev. D 68 (2003) 046001 [Erratum
ibid. D 70 (2004) 029903] [arXiv:hep-th/0303083].
[16] S. A. Abel and A. W. Owen, "N-point amplitudes in intersecting brane mod
els," Nucl. Phys. B 682 (2004) 183 [arXiv:hep-th/0310257].
[17] G. Aldazabal, S. Franco, L. E. Ibanez, R. Rabadan and A. M. Uranga, "D =
4 chiral string compactifications from intersecting branes," J. Math. Phys. 42
(2001) 3103 [arXiv:hep-th/0011073].
[18] D. Cremades, L. E. Ibanez and F. Marchesano, "Intersecting brane models of
particle physics and the Higgs mechanism," JHEP 0207 (2002) 022 [arXiv:hep
th/0203160].
[19] D. Cremades, L. E. Ibanez and F. Marchesano, "Yukawa couplings in inter
secting D-brane models," JHEP 0307 (2003) 038 [arXiv:hep-th/0302105].
Bibliography 128
[20] S. A. Abel and A. W. Owen, "Interactions in intersecting brane models," Nucl.
Phys. B 663 (2003) 197 [arXiv:hep-th/0303124].
[21] E. Witten, "D-branes and K-theory," JHEP 9812 (1998) 019 [arXiv:hep
th/9810188].
[22] F. G. Marchesano Buznego, "Intersecting D-brane models," arXiv:hep
th/0307252.
[23] F. Marchesano and G. Shiu, "Building MSSM flux vacua," JHEP 0411 (2004)
041 [arXiv:hep-th/0409132].
[24] R. Blumenhagen, M. Cvetic and T. Weigand, "Spacetime instanton corrections
in 4D string vacua- the seesaw mechanism for D-brane models," arXiv:hep
th/0609191.
[25] E. Witten, "World-sheet corrections via D-instantons," JHEP 0002 (2000)
030 [arXiv:hep-th/9907041].
[26] M. Billa, M. Frau, I. Pesando, F. Fucito, A. Lerda and A. Liccardo, JHEP
0302 (2003) 045 [arXiv:hep-th/0211250].
[27] N. Akerblom, R. Blumenhagen, D. Lust, E. Plauschinn and M. Schmidt
Sommerfeld, "Non-perturbative SQCD superpotentials from string instan
tons," arXiv:hep-th/0612132.
[28] S. A. Abel and M. D. Goodsell, "Realistic Yukawa couplings through instan
tons in intersecting brane worlds," arXiv:hep-th/0612110.
[29] N. Seiberg and E. Witten, "String theory and noncommutative geometry,"
JHEP 9909 (1999) 032 [hep-th/9908142].
[30] S. A. Abel and B. W. Schofield, "One-loop Yukawas on intersecting branes,"
JHEP 0506 (2005) 072 [arXiv:hep-th/0412206].
[31] S. A. Abel and M.D. Goodsell, "Intersecting brane worlds at one loop," JHEP
0602 (2006) 049 [arXiv:hep-th/0512072].
Bibliography 129
[32) D. Lust, P. Mayr, R. Richter and S. Stieberger, "Scattering of gauge, matter,
and moduli fields from intersecting branes," Nucl. Phys. B 696 (2004) 205
[arXiv:hep-th/0404134).
(33) M. Bertolini, M. Billo, A. Lerda, J. F. Morales and R. Russo, "Brane world
effective actions for D-branes with fluxes," [arXiv:hep-th/0512067).
(34) M. Berg, M. Haack and B. Kors, "String loop corrections to Kaehler potentials
in orientifolds," arXiv:hep-th/0508043.
(35) S. A. Abel, 0. Lebedev and J. Santiago, "Flavour in intersecting brane models
and bounds on the string scale," Nucl. Phys. B 696 (2004) 141 [arXiv:hep
ph/0312157].
(36] M. K. Gaillard and B. D. Nelson, "On Quadratic Divergences in Supergrav
ity, Vacuum Energy and the Supersymmetric Flavor Problem," arXiv:hep
ph/0511234.
(37] J. J. Atick, L. J. Dixon and A. Sen, "String Calculation of Fayet-Iliopoulos
D Terms in Arbitrary Supersymmetric Compactifications," Nucl.Phys. B292
109-149 (1987).
[38] S. Forste, G. Honecker and R. Schreyer, "Supersymmetric Z(N) x Z(M) ori
entifolds in 4D with D-branes at angles," Nucl. Phys. B 593 (2001) 127
[arXiv:hep-th/0008250].
[39) D. Lust and S. Stieberger, "Gauge threshold corrections in intersecting brane
world models," arXiv:hep-th/0302221.
(40) K. R. Dienes, E. Dudas and T. Gherghetta, "Grand unification at interme
diate mass scales through extra dimensions," Nucl. Phys. B 537 (1999) 47
[arXiv:hep-ph/9806292].
[41] I. Antoniadis, K. Benakli and A. Laugier, "Contact interactions in D-brane
models," JHEP 0105 (2001) 044 [arXiv:hep-th/0011281].
Bibliography 130
[42] S. A. Abel, M. Masip and J. Santiago, "Flavour changing neutral currents in
intersecting brane models," JHEP 0304 (2003) 057 [arXiv:hep-ph/0303087].
[43] E. Dudas and J. Mourad, "String theory predictions for future accelerators,"
Nucl. Phys. B 575 (2000) 3 [arXiv:hep-th/9911019].
[44] S. Cullen, M. Perelstein and M. E. Peskin, "TeV strings and collider probes
of large extra dimensions," Phys. Rev. D 62 (2000) 055012 [arXiv:hep
ph/0001166].
[45] N. Chamoun, S. Khalil and E. Lashin, "Fermion masses and mixing in in
tersecting branes scenarios," Phys. Rev. D 69 (2004) 095011 [arXiv:hep
ph/0309169].
[46] N. Kitazawa, T. Kobayashi, N. Maru and N. Okada, Eur. Phys. J. C 40, 579
(2005) [arXiv:hep-th/0406115].
[47] C. Kokorelis, "Standard model compactifications ofiiA Z(3) x Z(3) orientifolds
from intersecting D6-branes," Nucl. Phys. B 732, 341 (2006) [arXiv:hep
th/0412035].
[48] T. Higaki, N. Kitazawa, T. Kobayashi and K. j. Takahashi, "Flavor structure
and coupling selection rule from intersecting D-branes," Phys. Rev. D 72,
086003 (2005) [arXiv:hep-th/0504019].
[49] C. M. Chen, T. Li and D. V. Nanopoulos, "Standard-like model building on
type II orientifolds," Nucl. Phys. B 732, 224 (2006) [arXiv:hep-th/0509059].
[50] B. Dutta and Y. Mimura, "Properties of fermion mixings in intersecting D
brane models," Phys. Lett. B 633, 761 (2006) [arXiv:hep-ph/0512171].
[51] B. Dutta and Y. Mimura, "Lepton flavor violation in intersecting D-brane
models," Phys. Lett. B 638, 239 (2006) [arXiv:hep-ph/0604126].
[52] D. Cremades, L. E. Ibanez and F. Marchesano, "Computing Yukawa cou
plings from magnetized extra dimensions," JHEP 0405 (2004) 079 [arXiv:hep
th/0404229].
Bibliography 131
[53] L. E. Ibanez and A.M. Uranga, "Neutrino Majorana masses from string theory
instanton effects," arXiv:hep-th/0609213.
[54] M. Haack, D. Kreft, D. Lust, A. Van Proeyen and M. Zagermann, "Gaugino
condensates and D-terms from D7-branes," arXiv:hep-th/0609211.
[55] B. Florea, S. Kachru, J. McGreevy and N. Saulina, "Stringy instantons and
quiver gauge theories," arXiv:hep-th/0610003.
. ~ -. ''
[56] C. M. Chen, V. E. Mayes and D. V. Nanopoulos, "An MSSM-like Model from
Intersecting Branes on the z2 X z~ Orientifold," arXiv:hep-th/0612087.
[57] J. J. Atick, L. J. Dixon, P. A. Griffin and D. Nemeschansky, "Multiloop Twist
Field Correlation Functions For Z(N) Orbifolds," Nucl. Phys. B 298 (1988) 1.
[58] S. Abel, C. S. Chu and M. Goodsell, "Noncommutativity from the string
perspective: Modification of gravity at a mm without mm sized extra dimen
sions," arXiv:hep-th/0606248.
[59] M. R. Douglas and N. A. Nekrasov, "Noncommutative field theory," Rev.
Mod. Phys. 73 (2001) 977 [hep-th/0106048].
[60] R. J. Szabo, "Quantum field theory on noncommutative spaces," Phys. Rept.
378 (2003) 207 [hep-th/0109162].
[61] S. Minwalla, M. Van Raamsdonk and N. Seiberg, "Noncommutative pertur
bative dynamics," JHEP 0002, 020 (2000) [hep-th/9912072].
[62] A. Matusis, L. Susskind and N. Toumbas, "The IR/UV connection in the
non-commutative gauge theories," JHEP 0012 (2000) 002 [hep-th/0002075].
[63] A. Armoni, "Comments on perturbative dynamics of non-commutative Yang
Mills theory," Nucl. Phys. B 593 (2001) 229 [arXiv:hep-th/0005208].
[64] V. V. Khoze and G. Travaglini, "Wilsonian effective actions and the IR/UV
mixing in noncommutative gauge theories," JHEP 0101 (2001) 026 [hep
th/0011218].
Bibliography 132
[65] A. Armoni, R. Minasian and S. Theisen, Phys. Lett. B 513, 406 (2001)
[arXiv:hep-th/0102007].
[66] T. J. Hollowood, V. V. Khoze and G. Travaglini, "Exact results in noncom
mutative N = 2 supersymmetric gauge theories," JHEP 0105 (2001) 051 [hep
th/0102045].
[67] V. V. Khoze and J. Levell, "Noncommutative standard modelling," JHEP
0409 (2004) 019 [hep-th/0406178].
[68] L. Alvarez-Gaume and M. A. Vazquez-Mozo, "General properties of noncom
mutative field theories," Nucl. Phys. B 668 (2003) 293 [hep-th/0305093];
[69] L. Alvarez-Gaume and M.A. Vazquez-Mozo, "Comments on noncommutative
field theories," Springer Proc. Phys. 98 (2005) [arXiv:hep-th/0311244].
[70] C. S. Chu, K. Furuta and T. Inami, "Locality, causality and noncommutative
geometry," Int. J. Mod. Phys. A 21 (2006) 67 [arXiv:hep-th/0502012].
[71] J. Jaeckel, V. V. Khoze and A. Ringwald, "Telltale traces of U(1) fields in non
commutative standard model extensions," JHEP 0602 (2006) 028 [arXiv:hep
ph/0508075];
[72] S. A. Abel, J. Jaeckel, V. V. Khoze and A. Ringwald, "Photon defects in non
commutative standard model candidates," arXiv:hep-ph/0606106; S. A. Abel,
J. Jaeckel, V. V. Khoze and A. Ringwald, "Vacuum birefringence as a probe
of Planck scale noncommutativity," arXiv:hep-ph/0607188.
[73] J. Gomis, M. Kleban, T. Mehen, M. Rangamani and S. H. Shenker, "Noncom
mutative gauge dynamics from the string worldsheet," JHEP 0008 (2000) 011
[arXiv:hep-th/0003215].
[74] H. Liu and J. Michelson, "*-Trek: The one-loop N = 4 noncommutative SYM
action," Nucl. Phys. B 614 (2001) 279 [arXiv:hep-th/0008205].
Bibliography 133
[75] A. Connes, M. R. Douglas and A. Schwarz, "Noncommutative geometry and
matrix theory: Compactification on tori," JHEP 9802, 003 (1998) [arXiv:hep
th/9711162].
[76] M. R. Douglas and C. M. Hull, "D-branes and the noncommutative torus,"
JHEP 9802, 008 (1998) [arXiv:hep-th/9711165].
[77] C. S. Chu and P. M. Ho, "Noncommutative open string and D-brane," Nucl.
Phys. B 550, 151 (1999) [arXlv:hep-th/9812219].
[78] A. Bilal, C. S. Chu and R. Russo, "String theory and noncommutative field
theories at one loop," Nucl. Phys. B 582 (2000) 65 [arXiv:hep-th/0003180].
[79] C. S. Chu, R. Russo and S. Sciuto, "Multiloop string amplitudes with B
field and noncommutative QFT," Nucl. Phys. B 585 (2000) 193 [arXiv:hep
th/0004183].
[80] C. S. Chu, V. V. Khoze and G. Travaglini, "Dynamical breaking of super
symmetry in noncommutative gauge theories," Phys. Lett. B 513 (2001) 200
[arXiv:hep-th/0105187].
[81] H. Liu and J. Michelson, "Stretched strings in noncommutative field theory,"
Phys. Rev. D 62 (2000) 066003 [arXiv:hep-th/0004013].
[82] P. Di Vecchia, L. Magnea, A. Lerda, R. Russo and R. Marotta, "String tech
niques for the calculation of renormalization constants in field theory," Nucl.
Phys. B 469 (1996) 235 [arXiv:hep-th/9601143].
[83] A. Armoni and E. Lopez, "UV /IR mixing via closed strings and tachyonic
instabilities," Nucl. Phys. B 632 (2002) 240 [arXiv:hep-th/0110113].
[84] A. Armoni, E. Lopez and A. M. Uranga, "Closed strings tachyons and non
commutative instabilities," JHEP 0302 (2003) 020 [arXiv:hep-th/0301099].
[85] P. Di Vecchia, L. Magnea, A. Lerda, R. Marotta and R. Russo, "Two-loop
scalar diagrams from string theory," Phys. Lett. B 388 (1996) 65 [arXiv:hep
th/9607141].
Bibliography 134
[86] W. Fischler and L. Susskind, "Dilaton Tadpoles, String Condensates And Scale
Invariance. 2," Phys. Lett. B 173, 262 (1986).
[87] I. Antoniadis and C. Bachas, "Branes and the gauge hierarchy," Phys. Lett.
B 450, 83 (1999) [arXiv:hep-th/9812093].
[88] E. Dudas, G. Pradisi, M. Nicolosi and A. Sagnotti, "On tadpoles and vac
uum redefinitions in string theory," Nucl. Phys. B 708, 3 (2005) [arXiv:hep
th/0410101].
[89] S. A. Abel, J. Jaeckel, V. V. Khoze and A. Ringwald, "Noncommutativity,
extra dimensions, and power law running in the infrared," JHEP 0601 (2006)
105 [arXiv:hep-ph/0511197]
[90] L. Alvarez-Gaume, F. Meyer and M. A. Vazquez-Mozo, "Comments on non
commutative gravity," arXiv:hep-th/0605113.
[91] G. Aldazabal, L. E. Ibanez, F. Quevedo and A. M. Uranga, "D-branes at
singularities: A bottom-up approach to the string embedding of the standard
model," JHEP 0008, 002 (2000) [arXiv:hep-th/0005067];
[92] I. Antoniadis, C. Bachas, C. Fabre, H. Partouche and T. R. Taylor, "Aspects
of type I - type II - heterotic triality in four dimensions," Nucl. Phys. B 489
(1997) 160 [arXiv:hep-th/9608012].
[93] F. T. J. Epple, "Induced gravity on intersecting branes," JHEP 0409 (2004)
021 [arXiv:hep-th/0408105].
[94] V. A. Kostelecky and M. Mewes, "Cosmological constraints on Lorentz vio
lation in electrodynamics," Phys. Rev. Lett. 87, 251304 (2001) [arXiv:hep
ph/0111026].
[95] V. A. Kostelecky and M. Mewes, "Signals for Lorentz violation in electrody
namics," Phys. Rev. D 66, 056005 (2002) [arXiv:hep-ph/0205211].
[96] I. Mocioiu, M. Pospelov and R. Roiban, "Limits on the non-commutativity
scale," arXiv:hep-ph/0110011.
Bibliography 135
[97] I. Antoniadis, S. Dimopoulos and G. R. Dvali, "Millimeter range forces in
superstring theories with weak-scale compactification," Nucl. Phys. B 516, 70
(1998) [arXiv:hep-ph/9710204].
[98] I. Antoniadis, N. Arkani-Hamed, S. Dimopoulos and G. R. Dvali, "New di
mensions at a millimeter to a Fermi and superstrings at a TeV," Phys. Lett.
B 436, 257 (1998) [arXiv:hep-ph/9804398].
[99] G. R. Dvali, G. Gabadadze and M. Porrati, "4D gravity on a brane in 5D
Minkowski space," Phys. Lett. B 485, 208 (2000) [arXiv:hep-th/0005016].
[100] E. Kiritsis, N. Tetradis and T. N. Tomaras, "Thick branes and 4D gravity,"
JHEP 0108, 012 (2001) [arXiv:hep-th/0106050].
[101] C. D. Hoyle, U. Schmidt, B. R. Heckel, E. G. Adelberger, J. H. Gundlach,
D. J. Kapner and H. E. Swanson, "Sub-millimeter tests of the gravitational
inverse-square law: A search for 'large' extra dimensions," Phys. Rev. Lett. 86
(2001) 1418 [arXiv:hep-ph/0011014]; C. D. Hoyle, D. J. Kapner, B. R. Heckel,
E. G. Adelberger, J. H. Gundlach, U. Schmidt and H. E. Swanson, Phys. Rev.
D 70 (2004) 042004 [arXiv:hep-ph/0405262].
[102] F. R. Ruiz, "Gauge-fixing independence of IR divergences in non-commutative
U(1), perturbative tachyonic instabilities and supersymmetry," Phys. Lett. B
502 (2001) 274 [arXiv:hep-th/0012171].
[103] M. R. Gaberdiel, "D Branes from Conformal Field Theory", hep-th/0201113.
[104] P. Dorey, I. Runkel, R. Tateo and G. Watts, "g-function flow in perturbed
boundary conformal field theories," Nucl. Phys. B 578 (2000) 85 [arXiv:hep
th/9909216].
[105] J. J. Atick and A. Sen, "Covariant one loop fermion emission amplitudes in
closed string theories," Nucl. Phys. B 293 (1987) 317.